CN110298844A - X-ray contrastographic picture blood vessel segmentation and recognition methods and device - Google Patents
X-ray contrastographic picture blood vessel segmentation and recognition methods and device Download PDFInfo
- Publication number
- CN110298844A CN110298844A CN201910521211.2A CN201910521211A CN110298844A CN 110298844 A CN110298844 A CN 110298844A CN 201910521211 A CN201910521211 A CN 201910521211A CN 110298844 A CN110298844 A CN 110298844A
- Authority
- CN
- China
- Prior art keywords
- segmentation
- blood vessel
- decoder
- coronary angiography
- angiography image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- 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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10116—X-ray image
-
- 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
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/30101—Blood vessel; Artery; Vein; Vascular
Abstract
The present invention provides a kind of X-ray contrastographic picture blood vessel segmentation and recognition methods and device, and method includes: that coronary angiography image is input to blood vessel segmentation and identifies the encoder in network, exports the characteristic pattern of the coronary angiography image；The characteristic pattern of the encoder output is separately input into the blood vessel segmentation and identifies the segmentation decoder and identification decoder in network, the segmentation result that the coronary angiography image is exported by the segmentation decoder exports the recognition result of the coronary angiography image by the identification decoder；Wherein, the blood vessel segmentation and identification network are obtained after being trained based on coronary angiography image pattern, the segmentation result of predetermined each coronary angiography image pattern and predetermined vascular group label.The present invention can obtain more accurate identification and segmentation result using the blood vessel segmentation and identification network of training.
Description
Technical field
The invention belongs to technical field of image processing more particularly to a kind of X-ray contrastographic picture blood vessel segmentation and recognition methods
And device.
Background technique
Coronary angiography image is the common image of current diagnosis coronary heart disease, and doctor mainly infuses when observing coronary angiography image
Meaning power concentrates on several Main Branches of coronary artery, the lesions such as main detection is narrow, occlusion, thrombus and calcification.Therefore, accurately
Blood vessel segmentation and linear-elsatic buckling are very urgent.
Blood vessel identification is it is intended that the semantic information conventional method that image medium vessels structure division provides corresponding classification is mainly led to
It crosses based on tracking and model guidance (Model-Guided) algorithm and marks vessel branch classification.However these methods usually require
Corresponding 3D (Three-Dimensional, three-dimensional) vascular pattern, relies on manual intervention or needs artificial well-designed spy
Sign, can not cope with the overlapping of a variety of projection angles may cause in coronary angiography image blood vessel and intersect and the change of length of vessel
Change.
Deep learning method, especially CNN (Convolutional Neural Network, convolutional neural networks) are in recent years
To achieve breakthrough in coronary angiography image segmentation.The structure for generalling use class U-net is trained end-to-end, obtains two points
Class segmentation result.However these methods can face many challenges when handling blood vessel identification mission.Mainly due to 1) coronary angiography
Image projection angle difference bring complexity such as intersecting blood vessels and bifurcation problem etc.；2) the huge difference of different blood vessel structure bring
It is different；3) each classification pixel samples quantity of blood vessel is seriously uneven, it is difficult to optimize；4) background pixel and foreground vessel pixel is huge
Big quantity variance bring study is uneven.
In conclusion since the above problem causes when coronary angiography image is split and is identified, segmentation result and
Recognition result is inaccurate.
Summary of the invention
There are segmentation result and recognition results to overcome above-mentioned existing X-ray contrastographic picture blood vessel segmentation and recognition methods not
Accurate problem at least is partially solved the above problem, and the embodiment of the present invention provides a kind of X-ray contrastographic picture blood vessel segmentation
With recognition methods and device.
According to a first aspect of the embodiments of the present invention, a kind of X-ray contrastographic picture blood vessel segmentation and recognition methods, packet are provided
It includes:
Coronary angiography image is input to blood vessel segmentation and identifies the encoder in network, exports the coronary angiography image
Characteristic pattern；
The characteristic pattern of the encoder output is separately input into the blood vessel segmentation and identifies that the segmentation in network decodes
Device and identification decoder are exported the segmentation result of the coronary angiography image by the segmentation decoder, pass through the identification
Decoder exports the recognition result of the coronary angiography image；
Wherein, the blood vessel segmentation and identification network are made based on coronary angiography image pattern, predetermined each coronary artery
The segmentation result of shadow image pattern and predetermined vascular group label obtain after being trained.
According to a second aspect of the embodiments of the present invention, a kind of X-ray contrastographic picture blood vessel segmentation and identification device, packet are provided
It includes:
Coding module exports institute for coronary angiography image to be input to blood vessel segmentation and identifies the encoder in network
State the characteristic pattern of coronary angiography image；
Segmentation and identification module, for the characteristic pattern of the encoder output to be separately input into the blood vessel segmentation and is known
Segmentation decoder and identification decoder in other network, the segmentation of the coronary angiography image is exported by the segmentation decoder
As a result, exporting the recognition result of the coronary angiography image by the identification decoder；
Wherein, the blood vessel segmentation and identification network are made based on coronary angiography image pattern, predetermined each coronary artery
The segmentation result of shadow image pattern and predetermined vascular group label obtain after being trained.
In terms of third according to an embodiment of the present invention, also offer a kind of electronic equipment, including memory, processor and deposit
The computer program that can be run on a memory and on a processor is stored up, the processor calls described program instruction to be able to carry out
X-ray contrastographic picture blood vessel segmentation provided by any possible implementation in the various possible implementations of first aspect
With recognition methods.
The embodiment of the present invention provides a kind of X-ray contrastographic picture blood vessel segmentation and recognition methods and device, this method pass through by
The network of segmentation task and identification mission is integrated into a blood vessel segmentation and identification network, by the blood vessel segmentation and identification
Network carries out integration trainingt, so that segmentation task and identification mission supervision can be with collective effect in blood vessel segmentation and identification net
Network enables the blood vessel segmentation of training and identification network to obtain more accurate identification so that the network of two tasks promotes mutually
And segmentation result.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is this hair
Bright some embodiments for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is X-ray contrastographic picture blood vessel segmentation provided in an embodiment of the present invention and recognition methods overall flow schematic diagram；
Fig. 2 is X-ray contrastographic picture blood vessel segmentation provided in an embodiment of the present invention and the segmentation of recognition methods medium vessels and identifies
Schematic network structure；
Fig. 3 is electronic equipment overall structure diagram provided in an embodiment of the present invention.
Specific embodiment
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is this hair
Bright some embodiments for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
A kind of X-ray contrastographic picture blood vessel segmentation and recognition methods are provided in one embodiment of the invention, and Fig. 1 is this hair
The X-ray contrastographic picture blood vessel segmentation and recognition methods overall flow schematic diagram that bright embodiment provides, this method comprises: S101, it will
Coronary angiography image is input to blood vessel segmentation and identifies the encoder in network, exports the characteristic pattern of the coronary angiography image；
Wherein, coronary angiography image can be to pass through X-ray to shoot the image that coronarography is formed.Blood vessel segmentation with
Identification network for realizing the automatic segmentation and identification of coronary angiography image simultaneously, can based on U-net structure it is integrated divide with
It identifies two tasks, multi-task learning can be carried out in training.The encoder of i.e. two task sharings and the solution of segmentation task
Code device constitutes U-net structure, the decoder composition U-net structure of the encoder and identification mission.Segmentation task and identification mission
With decoder independent.U-net is a kind of deformation of convolutional neural networks, is obtained because its network structure is similar to letter U
Name.Using coronary angiography image as blood vessel segmentation and identification network input, can using continuous three frames coronary angiography image as
Input, to construct the blood vessel of multichannel input using two frame of front and back of an intermediate frame coronary angiography image as additional input
Segmentation and identification network.By the way of multichannel input, may learn the inner link between successive frame, promoted segmentation and
The precision of identification.The present embodiment is using coronary angiography image as the input of shared encoder, encoder output coronary angiography image
Characteristic pattern.
The characteristic pattern of the encoder output is separately input into the blood vessel segmentation and identifies the segmentation in network by S102
Decoder and identification decoder export the segmentation result of the coronary angiography image by the segmentation decoder, by described
Identification decoder exports the recognition result of the coronary angiography image；Wherein, the blood vessel segmentation and identification network are based on hat
The segmentation result and predetermined vascular group label of arteries and veins contrastographic picture sample, predetermined each coronary angiography image pattern
It is obtained after being trained.
Segmentation decoder is the decoder for the task that is split, and identification decoder is the decoder for carrying out identification mission.Point
It cuts task and learning tasks shares one and same coding device, the characteristic pattern of shared encoder output is separately input to two tasks
Decoder, travel to the information of coronary angiography image low level effectively in the decoder of two tasks.With U-net structure class
Seemingly, encoder is connected to the decoder of each task using jump.Encoder includes multiple down-sampling layers, divides the decoding of task
The quantity that layer is up-sampled in device and the decoder of identification mission is identical as the quantity of encoder down-sampling respectively, as shown in Fig. 2, compiling
Code device includes this 4 down-sampling layers of d1, d2, d3 and d4, and identification decoder and segmentation decoder respectively include 4 up-sampling layers.
In order to allow two tasks decoder learn how partial sharing, segmentation task decoder and identification mission decoder it
Between establish residual error connection, since segmentation task is a more basic task relative to identification mission, establish from segmentation
The decoder of task to identification mission decoder residual error connection, so that U-net is extended to residual error network structure.
When being trained to blood vessel segmentation and identification network, the entire blood vessel segmentation of integration trainingt and identification network join
It closes and the network structure of segmentation task and the network structure of identification mission is trained.Shared encoder table is arrived in study first
It reaches, then each task includes the decoder section of oneself, and learns how to merge respectively in blood vessel segmentation and identification network
From part.The network of two tasks is all made of intersection entropy loss and is trained, and dividing task and identification mission supervision can
With collective effect in blood vessel segmentation and identification network.The supervision of segmentation and the supervision of identification can all carry out about shared encoder
Beam.In addition, the supervision message of identification mission can also constrain the decoder of segmentation task, therefore two tasks can be mutual
Promote.
The present embodiment by the way that the network for dividing task and identification mission is integrated into a blood vessel segmentation and identification network,
By carrying out integration trainingt to the blood vessel segmentation and identification network, allow segmentation task and identification mission supervision common
Blood vessel segmentation and identification network are acted on, so that the network of two tasks promotes mutually, so that the blood vessel segmentation and identification of training
Network can obtain more accurate identification and segmentation result.
On the basis of the above embodiments, the coronary angiography image is exported by the identification decoder in the present embodiment
Recognition result the step of specifically include: for it is described identification decoder any layer, obtain this layer output characteristic pattern, and
There is the characteristic pattern of any layer output of the segmentation decoder of same scale with the characteristic pattern of this layer output；The layer is exported
Characteristic pattern and the segmentation decoder with same scale any layer output characteristic pattern as next layer of this layer
Input, until the recognition result of the next layer of output coronary angiography image of this layer.
Specifically, the present embodiment establishes the characteristic pattern of the segmentation task of same scale between the characteristic pattern of identification mission
It connects, the connection between characteristic pattern that the trapezoidal equivalent layer for indicating segmentation decoder and identification decoder in Fig. 2 exports.It will
The segmentation task of connection and the characteristic pattern of identification mission are established collectively as lower layer of decoder of identification of input, is on the one hand made
The supervision of identification mission to segmentation task decoder constrain, on the other hand fusion segmentation decoder output characteristic pattern into
Row identification, so that the feature that identification uses is more abundant, recognition result is more accurate.
On the basis of the various embodiments described above, the coronary angiography figure is exported by the identification decoder in the present embodiment
After the step of recognition result of picture further include: according to the recognition result of the coronary angiography image, generate each blood vessel mark
Sign corresponding probability graph.
Specifically, it is contemplated that a pixel in coronary angiography image may belong to multiple classifications, therefore to each classification
Export an individual probability graph, one blood vessel tag representation of each classification.To effectively solve intervascular overlapping and friendship
Fork problem detects accurate blood vessel structure in coronary angiography image.If in coronary angiography image medium vessels overlapping region
Pixel has multiple classifications, then these pixels are that multiclass intersects pixel or vascular bifurcation pixel.Decoder is identified as shown in Figure 2
Followed by the probability graph for classification each in recognition result, divide decoder followed by the probability for segmentation result
Figure.
On the basis of the above embodiments, the coronary angiography image is exported by the segmentation decoder in the present embodiment
Segmentation result, after the step of exporting the recognition result of the coronary angiography image by the identification decoder further include:
For any pixel in the segmentation result, the blood of the pixel of the position arest neighbors of the pixel is obtained from the recognition result
Pipe label obtains the recognition result of optimization using the blood vessel label of the pixel of the arest neighbors as the classification of the pixel.
Specifically, in order to which preferably using the detailed information of segmentation result, the present embodiment combination recognition result is tied with segmentation
Fruit carries out post-processing operation.Using the vessel segmentation of binaryzation as the basis of blood vessel structure, the blood vessel obtained for segmentation
Any pixel in structure obtains the classification of the pixel nearest apart from the location of pixels, i.e. blood vessel label in recognition result figure
As the classification of the pixel, thus the recognition result optimized, as shown in Figure 2.
On the basis of the various embodiments described above, coronary angiography image is being input to blood vessel segmentation and identification in the present embodiment
Encoder in network, before the step of exporting the characteristic pattern of the coronary angiography image further include: by coronary angiography image sample
Originally it is input to the encoder, exports the characteristic pattern of the coronary angiography image pattern；By the hat of the encoder output
The characteristic pattern of arteries and veins contrastographic picture sample is separately input into the segmentation decoder and the identification decoder, is solved by the segmentation
Code device exports the segmentation result of the coronary angiography image pattern, exports the coronary angiography image by the identification decoder
The recognition result of sample；The characteristic pattern for the coronary angiography image pattern that the identification decoder finally exports is gathered
Class obtains the cluster result of the characteristic pattern；According to segmentation result, the recognition result of the coronary angiography image pattern, and
The cluster result of the characteristic pattern is trained the blood vessel segmentation and identification network, until meeting preset termination condition.
Specifically, the present embodiment is added in identification mission differentiates cluster task, to realize the connectivity to blood vessel
Apply constraint.Differentiate that cluster task refers to each picture in the characteristic pattern for identifying the last one up-sampling layer output of decoder
Element is embedded into a higher dimensional space, and same category of pixel is flocked together, and different classes of pixel is separated.Such as Fig. 2
In, identification decoder includes this 4 up-sampling layers of u1, u2, u3 and u4, passes through the characteristic pattern of the last one up-sampling layer u4 output
Dypass export to obtain cluster result.In addition, being established in order to which the more acurrate acquisition blood vessel local context information of local message is added
The classification task of one local pixel block patch takes a certain size patch, by the class of its center pixel at random every time
Not Zuo Wei its classification markup information, establish classification task, obtain cluster result, so that use classes markup information is clustered to differentiating
It is constrained.When being trained to blood vessel segmentation and identification network, comprehensive cluster result, recognition result and segmentation result are to blood
Pipe segmentation is adjusted with the parameter in identification network, until meeting preset termination condition.
On the basis of the above embodiments, according to the segmentation result of the coronary angiography image pattern, knowledge in the present embodiment
Not as a result, and the characteristic pattern cluster result to the blood vessel segmentation and the objective function that is trained of identification network are as follows:
L=α Llab+βLsge+θLclu；
Lclu=γ1Ld+γ2Lp；
Wherein, L is target loss function, α, β, θ, γ1And γ2For weight coefficient, LlabFor the loss of the recognition result
Function, LsegFor the loss function of the segmentation result, LcluFor the loss function of the cluster result, LdFor the cluster result
Inter- object distance loss function, LpFor the loss function of the between class distance of the cluster result.
Specifically, in each iteration being trained to blood vessel segmentation and identification network, cluster result, segmentation knot are calculated
The loss of fruit and recognition result, feature of the loss between cluster result, segmentation result and recognition result and goldstandard image
Distance.Inter- object distance refers to that the characteristic distance between the pixel in a certain classification, between class distance refer to the pixel in a certain classification
With the characteristic distance between the pixel in other classifications.If inter- object distance and between class distance equivalence are important, γ1=γ2=1.
Llab、LsegAnd LcluTo intersect entropy loss.If three kinds of loss functions have same contribution degree, α=β=θ=1.Trained mesh
Be to keep the value of target loss function minimum.
A kind of X-ray contrastographic picture blood vessel segmentation and identification device, the device are provided in another embodiment of the present invention
For realizing the method in foregoing embodiments.Therefore, in each implementation of aforementioned X-ray contrastographic picture blood vessel segmentation and recognition methods
Description and definition in example, can be used for the understanding of each execution module in the embodiment of the present invention.The device includes coding module,
And segmentation and identification module, in which:
Coding module is used to for coronary angiography image to be input to blood vessel segmentation and identifies the encoder in network, described in output
The characteristic pattern of coronary angiography image；Segmentation is used to for the characteristic pattern of the encoder output being separately input into described with identification module
Blood vessel segmentation and segmentation decoder and identification decoder in identification network, export the coronary artery by the segmentation decoder and make
The segmentation result of shadow image exports the recognition result of the coronary angiography image by the identification decoder；Wherein, the blood
Pipe segmentation is the segmentation result based on coronary angiography image pattern, predetermined each coronary angiography image pattern with identification network
It is obtained after being trained with predetermined vascular group label.
The present embodiment provides a kind of electronic equipment, Fig. 3 is electronic equipment overall structure provided in an embodiment of the present invention signal
Figure, which includes: at least one processor 301, at least one processor 302 and bus 303；Wherein,
Processor 301 and memory 302 pass through bus 303 and complete mutual communication；
Memory 302 is stored with the program instruction that can be executed by processor 301, and the instruction of processor caller is able to carry out
Method provided by above-mentioned each method embodiment, for example, coronary angiography image is input to blood vessel segmentation and identification network
In encoder, export the characteristic pattern of the coronary angiography image；The characteristic pattern of the encoder output is separately input into institute
It states blood vessel segmentation and identifies the segmentation decoder and identification decoder in network, the coronary artery is exported by the segmentation decoder
The segmentation result of contrastographic picture exports the recognition result of the coronary angiography image by the identification decoder；Wherein, described
Blood vessel segmentation and identification network are the segmentation knots based on coronary angiography image pattern, predetermined each coronary angiography image pattern
Fruit and predetermined vascular group label obtain after being trained.
The present embodiment provides a kind of non-transient computer readable storage medium, non-transient computer readable storage medium storages
Computer instruction, computer instruction make computer execute method provided by above-mentioned each method embodiment, for example, by coronary artery
Contrastographic picture is input to blood vessel segmentation and identifies the encoder in network, exports the characteristic pattern of the coronary angiography image；By institute
The characteristic pattern for stating encoder output is separately input into the blood vessel segmentation and identifies that segmentation decoder and identification in network decode
Device is exported the segmentation result of the coronary angiography image by the segmentation decoder, exports institute by the identification decoder
State the recognition result of coronary angiography image；Wherein, the blood vessel segmentation and identification network are based on coronary angiography image pattern, in advance
The segmentation result and predetermined vascular group label of each coronary angiography image pattern first determined obtain after being trained.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above method embodiment can pass through
The relevant hardware of program instruction is completed, and program above-mentioned can be stored in a computer readable storage medium, the program
When being executed, step including the steps of the foregoing method embodiments is executed；And storage medium above-mentioned includes: ROM, RAM, magnetic disk or light
The various media that can store program code such as disk.
The apparatus embodiments described above are merely exemplary, wherein described, unit can as illustrated by the separation member
It is physically separated with being or may not be, component shown as a unit may or may not be physics list
Member, it can it is in one place, or may be distributed over multiple network units.It can be selected according to the actual needs
In some or all of the modules achieve the purpose of the solution of this embodiment.Those of ordinary skill in the art are not paying creativeness
Labour in the case where, it can understand and implement.
Through the above description of the embodiments, those skilled in the art can be understood that each embodiment can
It realizes by means of software and necessary general hardware platform, naturally it is also possible to pass through hardware.Based on this understanding, on
Stating technical solution, substantially the part that contributes to existing technology can be embodied in the form of software products in other words, should
Computer software product may be stored in a computer readable storage medium, such as ROM/RAM, magnetic disk, CD, including several fingers
It enables and using so that a computer equipment (can be personal computer, server or the network equipment etc.) executes each implementation
Method described in certain parts of example or embodiment.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify the technical solutions described in the foregoing embodiments or equivalent replacement of some of the technical features；
And these are modified or replaceed, technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution spirit and
Range.
Claims (10)
1. a kind of X-ray contrastographic picture blood vessel segmentation and recognition methods characterized by comprising
Coronary angiography image is input to blood vessel segmentation and identifies the encoder in network, exports the spy of the coronary angiography image
Sign figure；
By the characteristic pattern of the encoder output be separately input into the blood vessel segmentation and identify network in segmentation decoder and
It identifies decoder, the segmentation result of the coronary angiography image is exported by the segmentation decoder, is decoded by the identification
Device exports the recognition result of the coronary angiography image；
Wherein, the blood vessel segmentation and identification network are based on coronary angiography image pattern, predetermined each coronary angiography figure
Decent segmentation result and predetermined vascular group label obtain after being trained.
2. X-ray contrastographic picture blood vessel segmentation according to claim 1 and recognition methods, which is characterized in that the encoder
U-net structure is constituted with the segmentation decoder, the encoder and the identification decoder constitute U-net structure.
3. X-ray contrastographic picture blood vessel segmentation according to claim 1 and recognition methods, which is characterized in that from the segmentation
Decoder has residual error connection to identification decoder foundation.
4. X-ray contrastographic picture blood vessel segmentation according to claim 1 and recognition methods, which is characterized in that pass through the knowledge
Other decoder exports the step of recognition result of the coronary angiography image and specifically includes:
For any layer of the identification decoder, the characteristic pattern of this layer output is obtained, and is had with the characteristic pattern of this layer output
There is the characteristic pattern of any layer output of the segmentation decoder of same scale；
Using this layer output characteristic pattern and it is described with same scale segmentation decoder any layer output characteristic pattern as
Next layer of the input of this layer, until the recognition result of the next layer of output coronary angiography image of this layer.
5. X-ray contrastographic picture blood vessel segmentation according to claim 1 to 4 and recognition methods, which is characterized in that pass through
After the step of identification decoder exports the recognition result of the coronary angiography image further include:
According to the recognition result of the coronary angiography image, each corresponding probability graph of blood vessel label is generated.
6. X-ray contrastographic picture blood vessel segmentation according to claim 1 to 4 and recognition methods, which is characterized in that pass through
The segmentation decoder exports the segmentation result of the coronary angiography image, exports the coronary artery by the identification decoder and makes
After the step of recognition result of shadow image further include:
For any pixel in the segmentation result, the pixel of the position arest neighbors of the pixel is obtained from the recognition result
Blood vessel label obtain the recognition result of optimization using the blood vessel label of the pixel of the arest neighbors as the classification of the pixel.
7. X-ray contrastographic picture blood vessel segmentation according to claim 1 to 4 and recognition methods, which is characterized in that inciting somebody to action
Coronary angiography image is input to blood vessel segmentation and identifies the encoder in network, exports the characteristic pattern of the coronary angiography image
Before step further include:
Coronary angiography image pattern is input to the encoder, exports the characteristic pattern of the coronary angiography image pattern；
By the characteristic pattern of the coronary angiography image pattern of the encoder output be separately input into the segmentation decoder and
The identification decoder exports the segmentation result of the coronary angiography image pattern by the segmentation decoder, by described
Identification decoder exports the recognition result of the coronary angiography image pattern；
The characteristic pattern for the coronary angiography image pattern that the identification decoder finally exports is clustered, the spy is obtained
Levy the cluster result of figure；
According to the cluster result of the segmentation result of the coronary angiography image pattern, recognition result and the characteristic pattern to institute
It states blood vessel segmentation to be trained with identification network, until meeting preset termination condition.
8. X-ray contrastographic picture blood vessel segmentation according to claim 7 and recognition methods, which is characterized in that according to the hat
The cluster result of the segmentation result of arteries and veins contrastographic picture sample, recognition result and the characteristic pattern is to the blood vessel segmentation and knows
The objective function that other network is trained are as follows:
L=α Llab+βLsge+θLclu；
Lclu=γ1Ld+γ2Lp；
Wherein, L is target loss function, α, β, θ, γ1And γ2For weight coefficient, LlabFor the loss function of the recognition result,
LsegFor the loss function of the segmentation result, LcluFor the loss function of the cluster result, LdFor the class of the cluster result
The loss function of interior distance, LpFor the loss function of the between class distance of the cluster result.
9. a kind of X-ray contrastographic picture blood vessel segmentation and identification device characterized by comprising
Coding module exports the hat for coronary angiography image to be input to blood vessel segmentation and identifies the encoder in network
The characteristic pattern of arteries and veins contrastographic picture；
Segmentation and identification module, for the characteristic pattern of the encoder output to be separately input into the blood vessel segmentation and identification net
Segmentation decoder and identification decoder in network, the segmentation knot of the coronary angiography image is exported by the segmentation decoder
Fruit exports the recognition result of the coronary angiography image by the identification decoder；
Wherein, the blood vessel segmentation and identification network are based on coronary angiography image pattern, predetermined each coronary angiography figure
Decent segmentation result and predetermined vascular group label obtain after being trained.
10. a kind of electronic equipment including memory, processor and stores the calculating that can be run on a memory and on a processor
Machine program, which is characterized in that the processor realizes the X-ray radiography as described in any one of claim 1 to 8 when executing described program
The step of image blood vessel segmentation and recognition methods.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910521211.2A CN110298844B (en) | 2019-06-17 | 2019-06-17 | X-ray radiography image blood vessel segmentation and identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910521211.2A CN110298844B (en) | 2019-06-17 | 2019-06-17 | X-ray radiography image blood vessel segmentation and identification method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110298844A true CN110298844A (en) | 2019-10-01 |
| CN110298844B CN110298844B (en) | 2021-06-29 |
Family
ID=68027996
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910521211.2A Active CN110298844B (en) | 2019-06-17 | 2019-06-17 | X-ray radiography image blood vessel segmentation and identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110298844B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111008986A (en) * | 2019-11-20 | 2020-04-14 | 天津大学 | Remote sensing image segmentation method based on multitask semi-convolution |
| CN111047589A (en) * | 2019-12-30 | 2020-04-21 | 北京航空航天大学 | Attention-enhanced brain tumor auxiliary intelligent detection and identification method |
| CN111161240A (en) * | 2019-12-27 | 2020-05-15 | 上海联影智能医疗科技有限公司 | Blood vessel classification method, computer device and readable storage medium |
| CN111178420A (en) * | 2019-12-24 | 2020-05-19 | 北京理工大学 | Coronary segment labeling method and system on two-dimensional contrast image |
| CN111317462A (en) * | 2020-03-20 | 2020-06-23 | 佛山科学技术学院 | Blood flow imaging method and device based on U-net neural network |
| CN111612790A (en) * | 2020-04-29 | 2020-09-01 | 杭州电子科技大学 | Medical image segmentation method based on T-shaped attention structure |
| CN111784762A (en) * | 2020-06-01 | 2020-10-16 | 北京理工大学 | Method and device for extracting blood vessel center line of X-ray contrast image |
| CN113159050A (en) * | 2021-04-23 | 2021-07-23 | 广东工业大学 | Coronary artery CT image depth clustering and segmentation method and system |
| CN113408471A (en) * | 2021-07-02 | 2021-09-17 | 浙江传媒学院 | Non-green-curtain portrait real-time matting algorithm based on multitask deep learning |
| CN113902757A (en) * | 2021-10-09 | 2022-01-07 | 天津大学 | Blood vessel segmentation method based on self-attention mechanism and convolution neural network hybrid model |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2017101166A4 (en) * | 2017-08-25 | 2017-11-02 | Lai, Haodong MR | A Method For Real-Time Image Style Transfer Based On Conditional Generative Adversarial Networks |
| WO2018015414A1 (en) * | 2016-07-21 | 2018-01-25 | Siemens Healthcare Gmbh | Method and system for artificial intelligence based medical image segmentation |
| AU2018100325A4 (en) * | 2018-03-15 | 2018-04-26 | Nian, Xilai MR | A New Method For Fast Images And Videos Coloring By Using Conditional Generative Adversarial Networks |
| CN108550259A (en) * | 2018-04-19 | 2018-09-18 | 何澜 | Congestion in road judgment method, terminal device and computer readable storage medium |
| CN108764241A (en) * | 2018-04-20 | 2018-11-06 | 平安科技（深圳）有限公司 | Divide method, apparatus, computer equipment and the storage medium of near end of thighbone |
| CN109191476A (en) * | 2018-09-10 | 2019-01-11 | 重庆邮电大学 | The automatic segmentation of Biomedical Image based on U-net network structure |
| CN109389051A (en) * | 2018-09-20 | 2019-02-26 | 华南农业大学 | A kind of building remote sensing images recognition methods based on convolutional neural networks |
| CN109711411A (en) * | 2018-12-10 | 2019-05-03 | 浙江大学 | A kind of image segmentation and identification method based on capsule neuron |
-
2019
- 2019-06-17 CN CN201910521211.2A patent/CN110298844B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018015414A1 (en) * | 2016-07-21 | 2018-01-25 | Siemens Healthcare Gmbh | Method and system for artificial intelligence based medical image segmentation |
| AU2017101166A4 (en) * | 2017-08-25 | 2017-11-02 | Lai, Haodong MR | A Method For Real-Time Image Style Transfer Based On Conditional Generative Adversarial Networks |
| AU2018100325A4 (en) * | 2018-03-15 | 2018-04-26 | Nian, Xilai MR | A New Method For Fast Images And Videos Coloring By Using Conditional Generative Adversarial Networks |
| CN108550259A (en) * | 2018-04-19 | 2018-09-18 | 何澜 | Congestion in road judgment method, terminal device and computer readable storage medium |
| CN108764241A (en) * | 2018-04-20 | 2018-11-06 | 平安科技（深圳）有限公司 | Divide method, apparatus, computer equipment and the storage medium of near end of thighbone |
| CN109191476A (en) * | 2018-09-10 | 2019-01-11 | 重庆邮电大学 | The automatic segmentation of Biomedical Image based on U-net network structure |
| CN109389051A (en) * | 2018-09-20 | 2019-02-26 | 华南农业大学 | A kind of building remote sensing images recognition methods based on convolutional neural networks |
| CN109711411A (en) * | 2018-12-10 | 2019-05-03 | 浙江大学 | A kind of image segmentation and identification method based on capsule neuron |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111008986A (en) * | 2019-11-20 | 2020-04-14 | 天津大学 | Remote sensing image segmentation method based on multitask semi-convolution |
| CN111008986B (en) * | 2019-11-20 | 2023-09-05 | 天津大学 | Remote sensing image segmentation method based on multitasking semi-convolution |
| CN111178420B (en) * | 2019-12-24 | 2024-01-09 | 北京理工大学 | Coronary artery segment marking method and system on two-dimensional contrast image |
| CN111178420A (en) * | 2019-12-24 | 2020-05-19 | 北京理工大学 | Coronary segment labeling method and system on two-dimensional contrast image |
| CN111161240A (en) * | 2019-12-27 | 2020-05-15 | 上海联影智能医疗科技有限公司 | Blood vessel classification method, computer device and readable storage medium |
| CN111161240B (en) * | 2019-12-27 | 2024-03-05 | 上海联影智能医疗科技有限公司 | Blood vessel classification method, apparatus, computer device, and readable storage medium |
| CN111047589A (en) * | 2019-12-30 | 2020-04-21 | 北京航空航天大学 | Attention-enhanced brain tumor auxiliary intelligent detection and identification method |
| CN111047589B (en) * | 2019-12-30 | 2022-07-26 | 北京航空航天大学 | Attention-enhanced brain tumor auxiliary intelligent detection and identification method |
| CN111317462B (en) * | 2020-03-20 | 2023-11-03 | 佛山科学技术学院 | Blood flow imaging method and device based on U-net neural network |
| CN111317462A (en) * | 2020-03-20 | 2020-06-23 | 佛山科学技术学院 | Blood flow imaging method and device based on U-net neural network |
| CN111612790A (en) * | 2020-04-29 | 2020-09-01 | 杭州电子科技大学 | Medical image segmentation method based on T-shaped attention structure |
| CN111612790B (en) * | 2020-04-29 | 2023-10-17 | 杭州电子科技大学 | Medical image segmentation method based on T-shaped attention structure |
| CN111784762A (en) * | 2020-06-01 | 2020-10-16 | 北京理工大学 | Method and device for extracting blood vessel center line of X-ray contrast image |
| CN111784762B (en) * | 2020-06-01 | 2023-05-30 | 北京理工大学 | Method and device for extracting blood vessel center line of X-ray radiography image |
| CN113159050B (en) * | 2021-04-23 | 2023-03-10 | 广东工业大学 | Coronary artery CT image depth clustering and segmentation method and system |
| CN113159050A (en) * | 2021-04-23 | 2021-07-23 | 广东工业大学 | Coronary artery CT image depth clustering and segmentation method and system |
| CN113408471A (en) * | 2021-07-02 | 2021-09-17 | 浙江传媒学院 | Non-green-curtain portrait real-time matting algorithm based on multitask deep learning |
| CN113902757B (en) * | 2021-10-09 | 2022-09-02 | 天津大学 | Blood vessel segmentation method based on self-attention mechanism and convolution neural network hybrid model |
| CN113902757A (en) * | 2021-10-09 | 2022-01-07 | 天津大学 | Blood vessel segmentation method based on self-attention mechanism and convolution neural network hybrid model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110298844B (en) | 2021-06-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110298844A (en) | X-ray contrastographic picture blood vessel segmentation and recognition methods and device | |
| AU2016308097B2 (en) | Three-dimensional (3D) convolution with 3D batch normalization | |
| US20210065058A1 (en) | Method, apparatus, device and readable medium for transfer learning in machine learning | |
| CN109003679A (en) | A kind of cerebrovascular hemorrhage and ischemic prediction technique and device | |
| CN109325430A (en) | Real-time Activity recognition method and system | |
| CN109635812B (en) | The example dividing method and device of image | |
| CN111899244B (en) | Image segmentation method, network model training method, device and electronic equipment | |
| CN108764039A (en) | Building extracting method, medium and the computing device of neural network, remote sensing image | |
| CN109558892A (en) | A kind of target identification method neural network based and system | |
| CN113269737B (en) | Fundus retina artery and vein vessel diameter calculation method and system | |
| CN113554665A (en) | Blood vessel segmentation method and device | |
| CN109636788A (en) | A kind of CT image gall stone intelligent measurement model based on deep neural network | |
| CN112541924A (en) | Fundus image generation method, device, equipment and storage medium | |
| CN112529913A (en) | Image segmentation model training method, image processing method and device | |
| CN110163849A (en) | Training data processing method, disaggregated model training method and equipment | |
| CN109447096A (en) | A kind of pan path prediction technique and device based on machine learning | |
| CN111179269A (en) | PET image segmentation method based on multi-view and 3-dimensional convolution fusion strategy | |
| CN109658399A (en) | A kind of neck patch image-recognizing method and device | |
| CN111768467A (en) | Image filling method, device, equipment and storage medium | |
| CN110516572A (en) | A kind of method, electronic equipment and storage medium identifying competitive sports video clip | |
| Zhou et al. | Learning to address intra-segment misclassification in retinal imaging | |
| CN110009641A (en) | Crystalline lens dividing method, device and storage medium | |
| CN116188501B (en) | Medical image segmentation method based on multi-scale cross attention | |
| CN112364933A (en) | Image classification method and device, electronic equipment and storage medium | |
| CN114783597B (en) | Method and device for diagnosing multi-class diseases, electronic equipment and storage medium |
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 |