CN110298844B - X-ray radiography image blood vessel segmentation and identification method and device - Google Patents
X-ray radiography image blood vessel segmentation and identification method and device Download PDFInfo
- Publication number
- CN110298844B CN110298844B CN201910521211.2A CN201910521211A CN110298844B CN 110298844 B CN110298844 B CN 110298844B CN 201910521211 A CN201910521211 A CN 201910521211A CN 110298844 B CN110298844 B CN 110298844B
- Authority
- CN
- China
- Prior art keywords
- segmentation
- identification
- decoder
- coronary angiography
- result
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
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 invention provides a method and a device for segmenting and identifying blood vessels of an X-ray contrast image, wherein the method comprises the following steps: inputting a coronary angiography image into an encoder in a blood vessel segmentation and identification network, and outputting a characteristic diagram of the coronary angiography image; respectively inputting the feature map output by the encoder into a segmentation decoder and an identification decoder in the blood vessel segmentation and identification network, outputting the segmentation result of the coronary angiography image through the segmentation decoder, and outputting the identification result of the coronary angiography image through the identification decoder; the blood vessel segmentation and identification network is obtained by training based on the coronary angiography image samples, the predetermined segmentation results of the coronary angiography image samples and the predetermined blood vessel type labels. The invention can obtain more accurate recognition and segmentation results by using the trained blood vessel segmentation and recognition network.
Description
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to a method and a device for segmenting and identifying blood vessels of an X-ray contrast image.
Background
Coronary angiography images are commonly used for diagnosing coronary heart disease at present, and doctors mainly pay attention to a plurality of main branches of coronary artery when observing the coronary angiography images, and mainly observe lesions such as stenosis, occlusion, thrombus, calcification and the like. Therefore, accurate vessel segmentation and branch identification are highly urgent.
Conventional methods label vessel branch categories primarily by tracking and Model-Guided based algorithms. However, these methods usually require a corresponding 3D (Three-Dimensional) vessel model, rely on manual intervention or require manually elaborated features, and cannot cope with vessel overlap and intersection and vessel length changes that may be caused by multiple projection angles in the coronary angiography image.
In recent years, deep learning methods, particularly CNN (Convolutional Neural Network), have made a breakthrough in coronary angiography image segmentation. Usually, a structure similar to U-net is adopted for end-to-end training, and a binary classification segmentation result is obtained. However, these approaches face many challenges in dealing with the task of vessel identification. Mainly because of 1) the complexity caused by different projection angles of coronary angiography images, such as the problems of blood vessel crossing and bifurcation, etc.; 2) the great differences brought by different vascular structures; 3) the quantity of each category of pixel samples of the blood vessels is seriously unbalanced and is difficult to optimize; 4) the learning imbalance caused by the huge difference of the background pixels and the foreground vessel pixels.
In summary, due to the above-described problems, when segmenting and recognizing a coronary angiography image, the segmentation result and the recognition result are inaccurate.
Disclosure of Invention
In order to overcome or at least partially solve the above problems of inaccuracy of the segmentation result and the identification result of the existing X-ray contrast image vessel segmentation and identification method, embodiments of the present invention provide a method and an apparatus for segmenting and identifying a vessel of an X-ray contrast image.
According to a first aspect of the embodiments of the present invention, there is provided a method for segmenting and identifying a blood vessel in an X-ray contrast image, including:
inputting a coronary angiography image into an encoder in a blood vessel segmentation and identification network, and outputting a characteristic diagram of the coronary angiography image;
respectively inputting the feature map output by the encoder into a segmentation decoder and an identification decoder in the blood vessel segmentation and identification network, outputting the segmentation result of the coronary angiography image through the segmentation decoder, and outputting the identification result of the coronary angiography image through the identification decoder;
the blood vessel segmentation and identification network is obtained by training based on the coronary angiography image samples, the predetermined segmentation results of the coronary angiography image samples and the predetermined blood vessel type labels.
According to a second aspect of the embodiments of the present invention, there is provided an X-ray contrast image vessel segmentation and identification apparatus, including:
the encoding module is used for inputting the coronary angiography image into an encoder in a blood vessel segmentation and identification network and outputting a characteristic diagram of the coronary angiography image;
the segmentation and identification module is used for respectively inputting the feature map output by the encoder into a segmentation decoder and an identification decoder in the blood vessel segmentation and identification network, outputting the segmentation result of the coronary angiography image through the segmentation decoder, and outputting the identification result of the coronary angiography image through the identification decoder;
the blood vessel segmentation and identification network is obtained by training based on the coronary angiography image samples, the predetermined segmentation results of the coronary angiography image samples and the predetermined blood vessel type labels.
According to a third aspect of the embodiments of the present invention, there is also provided an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor invokes the program instructions to perform the method for segmenting and identifying a blood vessel in an X-ray contrast image provided in any one of the various possible implementations of the first aspect.
The embodiment of the invention provides a method and a device for segmenting and identifying blood vessels of an X-ray contrast image.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
Fig. 1 is a schematic overall flow chart of a method for segmenting and identifying blood vessels in an X-ray contrast image according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a blood vessel segmentation and identification network in the method for segmenting and identifying a blood vessel in an X-ray contrast image according to an embodiment of the present invention;
fig. 3 is a schematic view of an overall structure of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
In an embodiment of the present invention, a method for segmenting and identifying a blood vessel of an X-ray contrast image is provided, and fig. 1 is a schematic overall flow chart of the method for segmenting and identifying a blood vessel of an X-ray contrast image according to the embodiment of the present invention, where the method includes: s101, inputting a coronary angiography image into an encoder in a blood vessel segmentation and identification network, and outputting a characteristic diagram of the coronary angiography image;
the coronary angiography image may be an image formed by imaging coronary angiography with X-ray. The blood vessel segmentation and recognition network is used for realizing automatic segmentation and recognition of the coronary angiography image at the same time, can integrate two tasks of segmentation and recognition based on a U-net structure, and can perform multi-task learning during training. That is, an encoder shared by two tasks and a decoder for dividing the tasks form a U-net structure, and the encoder and the decoder for identifying the tasks form the U-net structure. The segmentation task and the recognition task have separate decoders. U-net is a variant of the convolutional neural network, named because its network structure is shaped like the letter U. The coronary angiography image is used as the input of the blood vessel segmentation and identification network, and three continuous frames of coronary angiography images can be used as the input, so that the front frame and the rear frame of the middle frame of coronary angiography image are used as additional input, and the blood vessel segmentation and identification network with multi-channel input is constructed. By adopting a multi-channel input mode, the internal relation between continuous frames can be learned, and the segmentation and identification precision is improved. The present embodiment takes a coronary angiography image as an input of a shared encoder, and the encoder outputs a feature map of the coronary angiography image.
S102, respectively inputting the feature map output by the encoder into a segmentation decoder and an identification decoder in the blood vessel segmentation and identification network, outputting the segmentation result of the coronary angiography image through the segmentation decoder, and outputting the identification result of the coronary angiography image through the identification decoder; the blood vessel segmentation and identification network is obtained by training based on the coronary angiography image samples, the predetermined segmentation results of the coronary angiography image samples and the predetermined blood vessel type labels.
The split decoder is a decoder performing a split task, and the recognition decoder is a decoder performing a recognition task. The segmentation task and the learning task share the same encoder, and feature maps output by the shared encoder are respectively input into decoders of the two tasks, so that low-level information of the coronary angiography image is effectively transmitted into the decoders of the two tasks. Similar to the U-net structure, the encoder is connected to the decoder of each task using hops. The encoder includes a plurality of down-sampling layers, and the number of up-sampling layers in the decoder of the division task and the decoder of the recognition task is the same as the number of down-sampling in the encoder, respectively, as shown in fig. 2, the encoder includes 4 down-sampling layers of d1, d2, d3, and d4, and the recognition decoder and the division decoder include 4 up-sampling layers, respectively. In order to let the decoders of both tasks learn how to share parts, a residual connection is established between the decoder of the splitting task and the decoder of the recognition task, and since the splitting task is a more fundamental task with respect to the recognition task, a residual connection is established from the decoder of the splitting task to the decoder of the recognition task, thereby extending the U-net into a residual network structure.
When the vessel segmentation and recognition network is trained, the whole vessel segmentation and recognition network is trained in an integrated manner, namely, the network structure of the segmentation task and the network structure of the recognition task are trained jointly. A shared encoder representation is first learned, then each task contains its own decoder part, and it is learned in the vessel segmentation and identification network how to fuse the respective parts. The networks of the two tasks are trained by adopting cross entropy loss, and supervision of the segmentation task and the recognition task can be commonly used in the blood vessel segmentation and recognition network. Both the supervision of segmentation and the supervision of identification impose constraints on the shared encoder. In addition, the supervisory information identifying the task also constrains the decoder that partitions the task, so that the two tasks can be facilitated by each other.
In the embodiment, the network of the segmentation task and the recognition task is integrated into a blood vessel segmentation and recognition network, and the blood vessel segmentation and recognition network is subjected to integrated training, so that supervision of the segmentation task and the recognition task can be commonly applied to the blood vessel segmentation and recognition network, the networks of the two tasks are mutually promoted, and the trained blood vessel segmentation and recognition network can obtain more accurate recognition and segmentation results.
On the basis of the foregoing embodiment, in this embodiment, the step of outputting the identification result of the coronary angiography image by the identification decoder specifically includes: for any layer of the identification decoder, acquiring a feature map output by the layer and a feature map output by any layer of the segmentation decoder with the same scale as the feature map output by the layer; and taking the feature map output by the layer and the feature map output by any layer of the segmentation decoder with the same scale as the input of the next layer of the layer until the next layer of the layer outputs the identification result of the coronary angiography image.
Specifically, the present embodiment establishes a connection between the feature maps of the segmentation task and the recognition task at the same scale, and the trapezoid in fig. 2 represents a connection between the feature maps output by the segmentation decoder and the corresponding layers of the recognition decoder. The segmentation task and the feature graph of the identification task which are connected are used as the input of the next layer of the identification decoder, so that the supervision of the identification task restricts the decoder of the segmentation task on one hand, and the feature graph output by the segmentation decoder is fused for identification on the other hand, so that the used features are richer in identification, and the identification result is more accurate.
On the basis of the foregoing embodiments, the step of outputting the identification result of the coronary angiography image by the identification decoder in this embodiment further includes: and generating a probability map corresponding to each blood vessel label according to the identification result of the coronary angiography image.
In particular, considering that a pixel in a coronary image may belong to multiple classes, a separate probability map is output for each class, each class being represented by a vessel label. Therefore, the problems of overlapping and crossing among blood vessels are effectively solved, and the accurate blood vessel structure in the coronary angiography image is detected. If the pixels of the blood vessel overlapping region in the coronary angiography image have multiple classes, the pixels are multiple classes of intersection pixels or blood vessel bifurcation pixels. The recognition decoder is immediately followed by a probability map for each class in the recognition result and the segmentation decoder is immediately followed by a probability map for the segmentation result as shown in fig. 2.
On the basis of the foregoing embodiment, in this embodiment, the step of outputting the segmentation result of the coronary angiography image by the segmentation decoder, and outputting the identification result of the coronary angiography image by the identification decoder further includes: and for any pixel in the segmentation result, acquiring the blood vessel label of the pixel with the nearest neighbor position of the pixel from the identification result, and acquiring the optimized identification result by taking the blood vessel label of the nearest neighbor pixel as the class of the pixel.
Specifically, in order to better utilize the detail information of the segmentation result, the present embodiment performs a post-processing operation in combination with the recognition result and the segmentation result. Using the binarized blood vessel segmentation result as the basis of the blood vessel structure, for any pixel in the obtained blood vessel structure by segmentation, obtaining the class of the pixel closest to the pixel position, namely, the blood vessel label, in the recognition result image as the class of the pixel, thereby obtaining the optimized recognition result, as shown in fig. 2.
On the basis of the foregoing embodiments, in this embodiment, before the step of inputting a coronary angiography image to an encoder in a blood vessel segmentation and identification network and outputting a feature map of the coronary angiography image, the method further includes: inputting a coronary angiography image sample into the encoder, and outputting a feature map of the coronary angiography image sample; inputting the feature maps of the coronary angiography image samples output by the encoder into the segmentation decoder and the identification decoder respectively, outputting the segmentation result of the coronary angiography image samples through the segmentation decoder, and outputting the identification result of the coronary angiography image samples through the identification decoder; clustering the feature map of the coronary angiography image sample finally output by the identification decoder to obtain a clustering result of the feature map; and training the blood vessel segmentation and recognition network according to the segmentation result and the recognition result of the coronary angiography image sample and the clustering result of the characteristic diagram until a preset termination condition is met.
Specifically, the distinguishing clustering task is added to the identification task, so that the constraint on the connectivity of the blood vessel is realized. The discriminant clustering task is to embed each pixel in a feature map output by a last upsampling layer of an identification decoder into a high-dimensional space, cluster the pixels of the same class together, and separate the pixels of different classes. As shown in fig. 2, the recognition decoder includes 4 upsampling layers u1, u2, u3 and u4, and the clustering result is obtained by side-path output of the feature map output by the last upsampling layer u 4. In addition, in order to add local information to more accurately acquire the local context information of the blood vessel, a classification task of a local pixel block patch is established, namely, a patch with a certain size is randomly selected each time, the class of the central pixel is used as class marking information, the classification task is established, a clustering result is acquired, and therefore the class marking information is used for constraining the discriminant clustering. When the vessel segmentation and recognition network is trained, the parameters in the vessel segmentation and recognition network are adjusted by integrating the clustering result, the recognition result and the segmentation result until a preset termination condition is met.
On the basis of the foregoing embodiment, in this embodiment, the objective function for training the blood vessel segmentation and recognition network according to the segmentation result and the recognition result of the coronary angiography image sample and the clustering result of the feature map is as follows:
L＝αLlab+βLsge+θLclu；
Lclu＝γ1Ld+γ2Lp；
wherein L is an objective loss function, alpha, beta, theta, gamma1And gamma2Is a weight coefficient, LlabAs a loss function of the recognition result, LsegAs a loss function of the segmentation result, LcluAs a loss function of the clustering result, LdA loss function of the intra-class distance of the clustering result, LpA loss function of the inter-class distance of the clustering result.
Specifically, in each iteration of training the blood vessel segmentation and recognition network, the loss of the clustering result, the segmentation result and the recognition result is calculated, and the loss is the characteristic distance between the clustering result, the segmentation result and the recognition result and the gold standard image. The intra-class distance refers to a characteristic distance between pixels in a certain class, and the inter-class distance refers to a characteristic distance between pixels in a certain class and pixels in other classes. If the intra-class distance and the inter-class distance are equally important, then γ1＝γ2＝1。Llab、LsegAnd LcluIs the cross entropy loss. If the three loss functions have the same degree of contribution, α ═ β ═ θ ═ 1. The goal of the training is to minimize the value of the objective loss function.
In another embodiment of the present invention, an X-ray contrast image blood vessel segmentation and identification device is provided, which is used for implementing the method in the foregoing embodiments. Therefore, the descriptions and definitions in the embodiments of the foregoing X-ray contrast image vessel segmentation and identification method can be used for understanding the various execution modules in the embodiments of the present invention. The device comprises an encoding module and a segmentation and identification module, wherein:
the encoding module is used for inputting the coronary angiography image into an encoder in a blood vessel segmentation and identification network and outputting a characteristic diagram of the coronary angiography image; the segmentation and identification module is used for respectively inputting the feature map output by the encoder into a segmentation decoder and an identification decoder in the blood vessel segmentation and identification network, outputting the segmentation result of the coronary angiography image through the segmentation decoder, and outputting the identification result of the coronary angiography image through the identification decoder; the blood vessel segmentation and identification network is obtained by training based on the coronary angiography image samples, the predetermined segmentation results of the coronary angiography image samples and the predetermined blood vessel type labels.
The embodiment provides an electronic device, and fig. 3 is a schematic view of an overall structure of the electronic device according to the embodiment of the present invention, where the electronic device includes: at least one processor 301, at least one memory 302, and a bus 303; wherein the content of the first and second substances,
the processor 301 and the memory 302 are communicated with each other through a bus 303;
the memory 302 stores program instructions executable by the processor 301, and the processor calls the program instructions to perform the methods provided by the above method embodiments, for example, the method includes: inputting a coronary angiography image into an encoder in a blood vessel segmentation and identification network, and outputting a characteristic diagram of the coronary angiography image; respectively inputting the feature map output by the encoder into a segmentation decoder and an identification decoder in the blood vessel segmentation and identification network, outputting the segmentation result of the coronary angiography image through the segmentation decoder, and outputting the identification result of the coronary angiography image through the identification decoder; the blood vessel segmentation and identification network is obtained by training based on the coronary angiography image samples, the predetermined segmentation results of the coronary angiography image samples and the predetermined blood vessel type labels.
The present embodiments provide a non-transitory computer-readable storage medium storing computer instructions that cause a computer to perform the methods provided by the above method embodiments, for example, including: inputting a coronary angiography image into an encoder in a blood vessel segmentation and identification network, and outputting a characteristic diagram of the coronary angiography image; respectively inputting the feature map output by the encoder into a segmentation decoder and an identification decoder in the blood vessel segmentation and identification network, outputting the segmentation result of the coronary angiography image through the segmentation decoder, and outputting the identification result of the coronary angiography image through the identification decoder; the blood vessel segmentation and identification network is obtained by training based on the coronary angiography image samples, the predetermined segmentation results of the coronary angiography image samples and the predetermined blood vessel type labels.
Those of ordinary skill in the art will understand that: all or part of the steps for implementing the method embodiments may be implemented by hardware related to program instructions, and the program may be stored in a computer readable storage medium, and when executed, the program performs the steps including the method embodiments; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (9)
1. A blood vessel segmentation and identification method for X-ray contrast images is characterized by comprising the following steps:
inputting a coronary angiography image into an encoder in a blood vessel segmentation and identification network, and outputting a characteristic diagram of the coronary angiography image;
respectively inputting the feature map output by the encoder into a segmentation decoder and an identification decoder in the blood vessel segmentation and identification network, outputting the segmentation result of the coronary angiography image through the segmentation decoder, and outputting the identification result of the coronary angiography image through the identification decoder;
the blood vessel segmentation and identification network is obtained by training based on coronary angiography image samples, predetermined segmentation results of the coronary angiography image samples and predetermined blood vessel type labels;
the step of outputting the identification result of the coronary angiography image through the identification decoder specifically includes:
for any layer of the identification decoder, acquiring a feature map output by the layer and a feature map output by any layer of the segmentation decoder with the same scale as the feature map output by the layer;
and taking the feature map output by the layer and the feature map output by any layer of the segmentation decoder with the same scale as the input of the next layer of the layer until the next layer of the layer outputs the identification result of the coronary angiography image.
2. The X-ray contrast image vessel segmentation and identification method according to claim 1, wherein the encoder and the segmentation decoder constitute a U-net structure, and the encoder and the identification decoder constitute a U-net structure.
3. The method of claim 1, wherein a residual connection is established from the segmentation decoder to the recognition decoder.
4. The method for segmenting and identifying blood vessels in X-ray contrast images according to any one of claims 1-3, wherein the step of outputting the identification result of the coronary contrast image by the identification decoder further comprises:
and generating a probability map corresponding to each blood vessel label according to the identification result of the coronary angiography image.
5. The method for segmenting and identifying blood vessels in X-ray contrast images according to any one of claims 1-3, wherein the step of outputting the segmentation result of the coronary contrast image by the segmentation decoder and outputting the identification result of the coronary contrast image by the identification decoder further comprises:
and for any pixel in the segmentation result, acquiring the blood vessel label of the pixel with the nearest neighbor position of the pixel from the identification result, and acquiring the optimized identification result by taking the blood vessel label of the nearest neighbor pixel as the class of the pixel.
6. The method for segmenting and identifying blood vessels of X-ray contrast images according to any one of claims 1-3, characterized in that before the step of inputting a coronary contrast image into an encoder in a blood vessel segmentation and identification network and outputting a feature map of the coronary contrast image, the method further comprises:
inputting a coronary angiography image sample into the encoder, and outputting a feature map of the coronary angiography image sample;
inputting the feature maps of the coronary angiography image samples output by the encoder into the segmentation decoder and the identification decoder respectively, outputting the segmentation result of the coronary angiography image samples through the segmentation decoder, and outputting the identification result of the coronary angiography image samples through the identification decoder;
clustering the feature map of the coronary angiography image sample finally output by the identification decoder to obtain a clustering result of the feature map;
and training the blood vessel segmentation and recognition network according to the segmentation result and the recognition result of the coronary angiography image sample and the clustering result of the characteristic diagram until a preset termination condition is met.
7. The method for segmenting and identifying blood vessels in X-ray contrast images according to claim 6, wherein the objective function for training the blood vessel segmentation and identification network according to the segmentation result, the identification result and the clustering result of the feature map of the coronary angiography image samples is as follows:
L＝αLlab+βLsge+θLclu；
Lclu＝γ1Ld+γ2Lp；
wherein L is an objective loss function, alpha, beta, theta, gamma1And gamma2Is a weight coefficient, LlabAs a loss function of the recognition result, LsegAs a loss function of the segmentation result, LcluAs a loss function of the clustering result, LdA loss function of the intra-class distance of the clustering result, LpA loss function of the inter-class distance of the clustering result.
8. An X-ray contrast image blood vessel segmentation and identification device is characterized by comprising:
the encoding module is used for inputting the coronary angiography image into an encoder in a blood vessel segmentation and identification network and outputting a characteristic diagram of the coronary angiography image;
the segmentation and identification module is used for respectively inputting the feature map output by the encoder into a segmentation decoder and an identification decoder in the blood vessel segmentation and identification network, outputting the segmentation result of the coronary angiography image through the segmentation decoder, and outputting the identification result of the coronary angiography image through the identification decoder;
the blood vessel segmentation and identification network is obtained by training based on coronary angiography image samples, predetermined segmentation results of the coronary angiography image samples and predetermined blood vessel type labels;
the step of outputting the identification result of the coronary angiography image through the identification decoder specifically includes:
for any layer of the identification decoder, acquiring a feature map output by the layer and a feature map output by any layer of the segmentation decoder with the same scale as the feature map output by the layer;
and taking the feature map output by the layer and the feature map output by any layer of the segmentation decoder with the same scale as the input of the next layer of the layer until the next layer of the layer outputs the identification result of the coronary angiography image.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program performs the steps of the method for vessel segmentation and identification of X-ray contrast images according to any one of claims 1 to 7.
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 CN110298844A (en) | 2019-10-01 |
| CN110298844B true 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) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| 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 |
| 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 |
| 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 |
| 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 |
| CN113408471B (en) * | 2021-07-02 | 2023-03-28 | 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 |
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 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110298844A (en) | 2019-10-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110298844B (en) | X-ray radiography image blood vessel segmentation and identification method and device | |
| CN110428432B (en) | Deep neural network algorithm for automatically segmenting colon gland image | |
| CN111899245A (en) | Image segmentation method, image segmentation device, model training method, model training device, electronic equipment and storage medium | |
| CN110490239B (en) | Training method, quality classification method, device and equipment of image quality control network | |
| CN111899244B (en) | Image segmentation method, network model training method, device and electronic equipment | |
| CN111696089A (en) | Arteriovenous determining method, device, equipment and storage medium | |
| US10997720B2 (en) | Medical image classification method and related device | |
| CN108764039A (en) | Building extracting method, medium and the computing device of neural network, remote sensing image | |
| CN110110723B (en) | Method and device for automatically extracting target area in image | |
| CN113205524B (en) | Blood vessel image segmentation method, device and equipment based on U-Net | |
| CN115063425B (en) | Reading knowledge graph-based structured inspection finding generation method and system | |
| CN111028248A (en) | Method and device for separating static and dynamic pulses based on CT (computed tomography) image | |
| CN113012155A (en) | Bone segmentation method in hip image, electronic device, and storage medium | |
| CN112884788B (en) | Cup optic disk segmentation method and imaging method based on rich context network | |
| CN115546570A (en) | Blood vessel image segmentation method and system based on three-dimensional depth network | |
| CN111178420B (en) | Coronary artery segment marking method and system on two-dimensional contrast image | |
| CN112541924A (en) | Fundus image generation method, device, equipment and storage medium | |
| CN115546231A (en) | Self-adaptive brain glioma segmentation method based on semi-supervised deep learning | |
| CN110363776B (en) | Image processing method and electronic device | |
| CN113469963B (en) | Pulmonary artery image segmentation method and device | |
| CN113763371A (en) | Pathological image cell nucleus segmentation method and device | |
| CN116188501B (en) | Medical image segmentation method based on multi-scale cross attention | |
| CN111915626A (en) | Automatic segmentation method and device for ventricle area of heart ultrasonic image and storage medium | |
| CN111209946A (en) | Three-dimensional image processing method, image processing model training method, and medium | |
| CN116664592A (en) | Image-based arteriovenous blood vessel separation method and device, electronic equipment and 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 |