CN112037186A - Coronary vessel extraction method and device based on multi-view model fusion - Google Patents
Coronary vessel extraction method and device based on multi-view model fusion Download PDFInfo
- Publication number
- CN112037186A CN112037186A CN202010855420.3A CN202010855420A CN112037186A CN 112037186 A CN112037186 A CN 112037186A CN 202010855420 A CN202010855420 A CN 202010855420A CN 112037186 A CN112037186 A CN 112037186A
- Authority
- CN
- China
- Prior art keywords
- coronary
- segmentation
- cta image
- heart
- original
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—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
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20132—Image cropping
-
- 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
- G06T2207/30048—Heart; Cardiac
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention provides a coronary artery blood vessel extraction method and device based on multi-view model fusion, wherein the method comprises the following steps: inputting an original coronary artery CTA image to obtain a heart segmentation result, and cutting the original CTA image according to the size of the heart to obtain a cut coronary artery CTA image; training a coronary vessel segmentation network by using a multi-view image, wherein the coronary vessel segmentation network comprises a first coronary vessel segmentation model, a second coronary vessel segmentation model and a third coronary vessel segmentation model; the method comprises the steps of utilizing a first coronary artery blood vessel segmentation model to segment a cut coronary artery CTA image to obtain a first segmentation result, utilizing a second coronary artery blood vessel segmentation model to segment the cut coronary artery CTA image to obtain a second segmentation result, utilizing a third coronary artery blood vessel segmentation model to segment the cut coronary artery CTA image to obtain a third segmentation result, and combining the first segmentation result, the second segmentation result and the third segmentation result to obtain a final coronary artery blood vessel segmentation result.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a coronary artery blood vessel extraction method and device based on multi-view model fusion.
Background
The angiography technology is widely applied to clinical diagnosis and treatment, and the segmentation algorithm can realize automatic blood vessel reconstruction (such as head and neck blood vessels, coronary arteries and the like), so that the working pressure of technicians is reduced, and the operating efficiency of hospitals is greatly improved. In an actual scene, some external factors (such as artifacts, noise, shooting technology and the like) can affect the quality of blood vessel imaging, a CT image is analyzed from a single view such as an axial position, and the trend of some blood vessels is not easy to judge, so that the learning needs to be performed by combining a plurality of views, and the segmentation precision is improved.
However, different coronary images have different scanning fields of view, and the whole image is used for segmenting the coronary blood vessels, so that calculation of many regions is wasted. When the coronary vessel is segmented, the single-view learning capability is limited, and multiple views can utilize more information, thereby being more beneficial to improving the segmentation effect.
Disclosure of Invention
The present invention aims to provide a coronary vessel extraction method and apparatus based on multi-view model fusion that overcomes or at least partially solves the above mentioned problems.
In order to achieve the purpose, the technical scheme of the invention is realized as follows:
one aspect of the present invention provides a coronary artery blood vessel extraction method based on multi-view model fusion, including: inputting an original coronary artery CTA image, segmenting a heart to obtain a heart segmentation result, and cutting the original CTA image according to the size of the heart to obtain a cut coronary artery CTA image; training a coronary vessel segmentation network by using the multi-view image, wherein the coronary vessel segmentation network comprises a first coronary vessel segmentation model, a second coronary vessel segmentation model and a third coronary vessel segmentation model; the method comprises the steps of utilizing a first coronary artery blood vessel segmentation model to segment a cut coronary artery CTA image to obtain a first segmentation result, utilizing a second coronary artery blood vessel segmentation model to segment the cut coronary artery CTA image to obtain a second segmentation result, utilizing a third coronary artery blood vessel segmentation model to segment the cut coronary artery CTA image to obtain a third segmentation result, and combining the first segmentation result, the second segmentation result and the third segmentation result to obtain a final coronary artery blood vessel segmentation result.
The method for obtaining the subimages for coronary vessel segmentation by clipping the original CTA image according to the size of the heart comprises the following steps: the maximum connected domain of the heart segmentation result is reserved, the initial bounding box of the maximum connected domain is calculated, and the coordinates of the initial bounding box are (x)0，x1，y0，y1，z0，z1)，x0，x1Denotes its coordinate range on the x-axis, y0，y1Denotes its coordinate range on the y-axis, z0，z1Representing its coordinate range on the z-axis; in the case of the original coronary CTA image size of (W, H, D), the original bounding box is spread by a certain physical distance around, (W, H, D), and the coordinates of the new bounding box are determined: x'0＝max(0，x0-w/s0)；x′1＝min(W，x1+w/s0)；y′0＝max(0，y0-h/s1)；y′1＝min(H，y1+ h/s1)；z′0＝max(0，z0-d/s2)；z′1＝min(D，z1+d/s2) (ii) a Wherein(s)0，s1，s2) Spacing information for the original coronary CTA image, which includes: pixel pitch and layer pitch; and clipping the original coronary CTA image by using the new bounding box to obtain the clipped coronary CTA image.
The method comprises the following steps of inputting an original coronary artery CTA image, segmenting a heart, and obtaining a heart segmentation result, wherein the heart segmentation result comprises the following steps: and inputting the original coronary artery CTA image into a CNN network, and segmenting the heart to obtain a heart segmentation result.
Wherein, training the coronary vessel segmentation network by using the multi-view image comprises the following steps: taking a slice block along the axial position to train a first coronary vessel segmentation model; taking a slice block along the sagittal position to train a second coronary vessel segmentation model; and taking a slice block along the coronal plane to train a third coronary vessel segmentation model.
The invention provides a coronary vessel extraction device based on multi-view model fusion, which comprises: the cutting module is used for inputting an original coronary artery CTA image, segmenting the heart to obtain a heart segmentation result, and cutting the original CTA image according to the size of the heart to obtain a cut coronary artery CTA image; the training module is used for training a coronary artery segmentation network by utilizing the multi-view image, and the coronary artery segmentation network comprises a first coronary artery segmentation model, a second coronary artery segmentation model and a third coronary artery segmentation model; and the segmentation module is used for segmenting the clipped coronary artery CTA image by using the first coronary artery blood vessel segmentation model to obtain a first segmentation result, segmenting the clipped coronary artery CTA image by using the second coronary artery blood vessel segmentation model to obtain a second segmentation result, segmenting the clipped coronary artery CTA image by using the third coronary artery blood vessel segmentation model to obtain a third segmentation result, and merging the first segmentation result, the second segmentation result and the third segmentation result to obtain a final coronary artery blood vessel segmentation result.
The clipping module clips the original CTA image according to the size of the heart in the following way to obtain a sub-image for coronary vessel segmentation: a clipping module, specifically for preserving the maximum connected domain of the heart segmentation result, and calculating the initial bounding box of the maximum connected domain, the coordinates of the initial bounding box being (x)0，x1，y0，y1，z0，z1)，x0，x1Denotes its coordinate range on the x-axis, y0，y1Denotes its coordinate range on the y-axis, z0，z1Representing its coordinate range on the z-axis; in the case of the original coronary CTA image size of (W, H, D), the original bounding box is spread by a certain physical distance around, (W, H, D), and the coordinates of the new bounding box are determined: x'0＝max(0，x0-w/s0)；x′1＝min(W，x1+w/s0)； y′0＝max(0，y0-h/s1)；y′1＝min(H，y1+h/s1)；z′0＝max(0，z0-d/s2)；z′1＝min(D，z1+ d/s2) (ii) a Wherein(s)0，s1，s2) Spacing information for the original coronary CTA image, which includes: pixel pitch and layer pitch; and clipping the original coronary CTA image by using the new bounding box to obtain the clipped coronary CTA image.
The clipping module inputs an original coronary artery CTA image in the following mode to segment the heart, and a heart segmentation result is obtained: the clipping module is specifically used for inputting the original coronary artery CTA image into a CNN network to segment the heart and obtain a heart segmentation result.
The training module trains the coronary vessel segmentation network by using the multi-view image in the following way: the training module is specifically used for taking a slice block along the axial position to train a first coronary vessel segmentation model; taking a slice block along the sagittal position to train a second coronary vessel segmentation model; and taking a slice block along the coronal plane to train a third coronary vessel segmentation model.
Therefore, by the coronary artery blood vessel extraction method and device based on multi-view model fusion, provided by the invention, the image space is cut by utilizing the heart structure, so that the network is more concentrated in learning the coronary artery blood vessel distribution area, and the problem that different coronary artery image scanning views are different is solved. The trend information of the blood vessel is better learned by using multiple views, and accurate coronary vessel segmentation is facilitated.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on the drawings without creative efforts.
Fig. 1 is a flowchart of a coronary artery extraction method based on multi-view model fusion according to an embodiment of the present invention;
fig. 2 is a schematic frame diagram of a coronary artery extraction method based on multi-view model fusion according to an embodiment of the present invention;
FIG. 3 is an original image and a cropped image provided by an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a coronary artery extraction device based on multi-view model fusion according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
The method for extracting coronary artery blood vessels based on multi-view model fusion provided by the embodiment of the invention is described below with reference to fig. 1 and 2, and with reference to fig. 1 and 2, the method for extracting coronary artery blood vessels based on multi-view model fusion provided by the embodiment of the invention comprises:
s1, inputting the original coronary CTA image, dividing the heart to obtain the heart division result, and cutting the original CTA image according to the size of the heart to obtain the cut coronary CTA image.
Specifically, an original coronary CTA image is input, and then a mask of the heart is segmented, wherein optionally, the heart is segmented by using a CNN network, and then the original coronary CTA image is clipped according to the size of the heart to obtain a sub-image for coronary vessel segmentation.
As an alternative to the embodiments of the present invention, cropping the original CTA image according to the size of the heart, and obtaining sub-images for coronary vessel segmentation, comprises: the maximum connected domain of the heart segmentation result is reserved, the initial bounding box of the maximum connected domain is calculated, and the coordinates of the initial bounding box are (x)0，x1，y0，y1，z0，z1)，x0，x1Representing its coordinate range on the x-axis, y0，y1denotes its coordinate range on the y-axis, z0，z1Representing its coordinate range on the z-axis; in the case of the original coronary CTA image size of (W, H, D), the original bounding box is spread by a certain physical distance around, (W, H, D), and the coordinates of the new bounding box are determined: x'0＝max(0，x0-w/s0)；x′1＝min(W，x1+w/s0)；y′0＝max(0，y0- h/s1)；y′1＝min(H，y1+h/s1)；z′0＝max(0，z0-d/s2)；z′1＝min(D，z1+d/s2) (ii) a Wherein(s)0，s1，s2) Spacing information for the original coronary CTA image, which includes: pixel pitch and layer pitch; and clipping the original coronary CTA image by using the new bounding box to obtain the clipped coronary CTA image.
Specifically, the maximum connected domain of the heart segmentation result is retained, and its initial bounding box B0 is calculated, the coordinates of which are (x)0，x1，y0，y1，z0，z1)，x0，x1Denotes its coordinate range on the x-axis, y0，y1Denotes its coordinate range on the y-axis, z0，z1Representing its coordinate range on the z-axis. Assuming that the original CTA image is of size (W, H, D) and extends a certain physical distance around the original bounding box, (W, H, D), the new bounding box B1 has the following coordinate ranges:
x′0＝max(0，x0-w/s0)
x′1＝min(W，x1+w/s0)
y′0＝max(0，y0-h/s1)
y′1＝min(H，y1+h/s1)
z′0＝max(0，z0-d/s2)
z′1＝min(D，z1+d/s2)
wherein(s)0，s1，s2) Spacing information (pixel pitch and layer pitch, typically in millimeters) of the original coronary CTA image. Typically (w, h, d) is between 1 cm and 3 cm.
The original CTA is clipped by B1 to obtain a coronary blood vessel region to be segmented with a uniform visual field, as shown in fig. 3.
S2, training a coronary artery segmentation network by using the multi-view image, wherein the coronary artery segmentation network comprises a first coronary artery segmentation model, a second coronary artery segmentation model and a third coronary artery segmentation model.
Specifically, a clipped coronary CTA image is obtained, and then a coronary vessel segmentation network is trained by using a multi-view image.
As an optional implementation manner of the embodiment of the present invention, the training of the coronary artery segmentation network by using the multi-view image includes: taking a slice block along the axial position to train a first coronary vessel segmentation model; taking a slice block along the sagittal position to train a second coronary vessel segmentation model; and taking a slice block along the coronal plane to train a third coronary vessel segmentation model.
Specifically, a slice block (continuous k-layer slice) is taken along the axial robust position to train a model A; taking a slice block training model S along the vector bit; taking a slice block training model C along the coronal bit; each view trains a coronary segmentation model.
As an alternative to the embodiment of the present invention, in addition to training the model with three views, i.e. axial, coronal and sagittal, as in the present invention, if the space is considered as a sphere, more views can be obtained by projection along multiple directions.
S3, the first coronary artery blood vessel segmentation model is used for segmenting the clipped coronary artery CTA image to obtain a first segmentation result, the second coronary artery blood vessel segmentation model is used for segmenting the clipped coronary artery CTA image to obtain a second segmentation result, the third coronary artery blood vessel segmentation model is used for segmenting the clipped coronary artery CTA image to obtain a third segmentation result, and the first segmentation result, the second segmentation result and the third segmentation result are combined to obtain a final coronary artery blood vessel segmentation result.
Specifically, for the clipped coronary CTA image, a segmentation result SA is obtained using the model a, a segmentation result SS is obtained using the model S, and a segmentation result SC is obtained using the model C. And finally, combining the SA, the SS and the SC to obtain a final coronary vessel segmentation result.
Therefore, the coronary artery blood vessel extraction method based on multi-view model fusion provided by the embodiment of the invention standardizes the coronary artery CTA images of different visual fields by using the heart, trains a plurality of coronary artery segmentation networks by using the multi-view images, and fuses the results of the multi-view segmentation networks, thereby realizing the extraction of the coronary artery blood vessel. Standardizing coronary artery CTA images with different visual fields to obtain sub-images with more consistent visual field range, and reducing the influence and calculation of irrelevant areas of blood vessel segmentation; the trend of coronary vessels is learned by using the multi-view images, and accurate prediction of the vessels is facilitated.
Fig. 4 is a schematic structural diagram of a coronary artery extraction device based on multi-view model fusion according to an embodiment of the present invention, in which the above method is applied, and the structure of the coronary artery extraction device based on multi-view model fusion is only briefly described below, and other things are not at all, please refer to the related description in the above coronary artery extraction method based on multi-view model fusion, see fig. 4, and the coronary artery extraction device based on multi-view model fusion according to an embodiment of the present invention includes:
the cutting module is used for inputting an original coronary artery CTA image, segmenting the heart to obtain a heart segmentation result, and cutting the original CTA image according to the size of the heart to obtain a cut coronary artery CTA image;
the training module is used for training a coronary artery segmentation network by utilizing the multi-view image, and the coronary artery segmentation network comprises a first coronary artery segmentation model, a second coronary artery segmentation model and a third coronary artery segmentation model;
and the segmentation module is used for segmenting the clipped coronary artery CTA image by using the first coronary artery blood vessel segmentation model to obtain a first segmentation result, segmenting the clipped coronary artery CTA image by using the second coronary artery blood vessel segmentation model to obtain a second segmentation result, segmenting the clipped coronary artery CTA image by using the third coronary artery blood vessel segmentation model to obtain a third segmentation result, and merging the first segmentation result, the second segmentation result and the third segmentation result to obtain a final coronary artery blood vessel segmentation result.
As an alternative implementation of the embodiment of the present invention, the cropping module crops the original CTA image according to the size of the heart to obtain sub-images for coronary vessel segmentation by: a clipping module, specifically for preserving the maximum connected domain of the heart segmentation result, and calculating the initial bounding box of the maximum connected domain, the coordinates of the initial bounding box being (x)0，x1，y0，y1，z0，z1)，x0，x1Denotes its coordinate range on the x-axis, y0，y1Denotes its coordinate range on the y-axis, z0，z1Representing its coordinate range on the z-axis; in the case of the original coronary CTA image size of (W, H, D), the original bounding box is spread by a certain physical distance around, (W, H, D), and the coordinates of the new bounding box are determined: x'0＝ max(0，x0-w/s0)；x′1＝min(W，x1+w/s0)；y′0＝max(0，y0-h/s1)；y′1＝min(H，y1+ h/s1)；z′0＝max(0，z0-d/s2)；z′1＝min(D，z1+d/s2) (ii) a Wherein(s)0，s1，s2) Spacing information for the original coronary CTA image, which includes: pixel pitch and layer pitch; and clipping the original coronary CTA image by using the new bounding box to obtain the clipped coronary CTA image.
As an alternative implementation of the embodiment of the present invention, the clipping module inputs the original coronary CTA image to segment the heart to obtain the heart segmentation result as follows: the clipping module is specifically used for inputting the original coronary artery CTA image into a CNN network to segment the heart and obtain a heart segmentation result.
As an optional implementation of the embodiment of the present invention, the training module trains the coronary vessel segmentation network by using the multi-view image in the following manner: the training module is specifically used for taking a slice block along the axial position to train a first coronary vessel segmentation model; taking a slice block along the sagittal position to train a second coronary vessel segmentation model; and taking a slice block along the coronal plane to train a third coronary vessel segmentation model.
Therefore, the coronary artery blood vessel extraction device based on multi-view model fusion provided by the embodiment of the invention standardizes the coronary artery CTA images of different visual fields by using the heart, trains a plurality of coronary artery segmentation networks by using the multi-view images, and fuses the results of the multi-view segmentation networks, thereby realizing the extraction of the coronary artery blood vessel. Standardizing coronary artery CTA images with different visual fields to obtain sub-images with more consistent visual field range, and reducing the influence and calculation of irrelevant areas of blood vessel segmentation; the trend of coronary vessels is learned by using the multi-view images, and accurate prediction of the vessels is facilitated.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims (8)
1. A coronary vessel extraction method based on multi-view model fusion is characterized by comprising the following steps:
inputting an original coronary artery CTA image, segmenting a heart to obtain a heart segmentation result, and cutting the original CTA image according to the size of the heart to obtain a cut coronary artery CTA image;
training a coronary vessel segmentation network by using a multi-view image, wherein the coronary vessel segmentation network comprises a first coronary vessel segmentation model, a second coronary vessel segmentation model and a third coronary vessel segmentation model;
the first coronary artery blood vessel segmentation model is used for segmenting the clipped coronary artery CTA image to obtain a first segmentation result, the second coronary artery blood vessel segmentation model is used for segmenting the clipped coronary artery CTA image to obtain a second segmentation result, the third coronary artery blood vessel segmentation model is used for segmenting the clipped coronary artery CTA image to obtain a third segmentation result, and the first segmentation result, the second segmentation result and the third segmentation result are combined to obtain a final coronary artery blood vessel segmentation result.
2. The method of claim 1, wherein said cropping the original CTA image according to the size of the heart, resulting in sub-images for coronary vessel segmentation, comprises:
preserving the maximum connected domain of the heart segmentation result, and calculating an initial bounding box of the maximum connected domain, wherein the coordinates of the initial bounding box are (x)0，x1，y0，y1，z0，z1)，x0，x1Denotes its coordinate range on the x-axis, y0，y1Denotes its coordinate range on the y-axis, z0，z1Representing its coordinate range on the z-axis;
in the case where the original coronary CTA image is of size (W, H, D), the original bounding box is spread around by a physical distance, (W, H, D), and the coordinates of the new bounding box are determined:
x′0＝max(0，x0-w/s0)；
x′1＝min(W，x1+w/s0)；
y′0＝max(0，y0-h/s1)；
y′1＝min(H，y1+h/s1)；
z′0＝max(0，z0-d/s2)；
z′1＝min(D，z1+d/s2)；
wherein(s)0，s1，s2) Spacing information for the original coronary CTA image, the spacing information comprising: pixel pitch and layer pitch;
and clipping the original coronary CTA image by using the new bounding box to obtain a clipped coronary CTA image.
3. The method of claim 1, wherein the inputting of the original coronary CTA image, segmenting the heart, and obtaining the heart segmentation result comprises:
and inputting the original coronary CTA image into a CNN network, and segmenting the heart to obtain the heart segmentation result.
4. The method of claim 1, wherein training the coronary vessel segmentation network using the multi-view image comprises:
taking a slice block along the axial position to train the first coronary vessel segmentation model;
taking a slice block along a sagittal position to train the second coronary vessel segmentation model;
and taking a slice block along the coronary position to train the third coronary vessel segmentation model.
5. A coronary vessel extraction device based on multi-view model fusion is characterized by comprising:
the cutting module is used for inputting an original coronary artery CTA image, segmenting the heart to obtain a heart segmentation result, and cutting the original CTA image according to the size of the heart to obtain a cut coronary artery CTA image;
the training module is used for training a coronary artery segmentation network by utilizing the multi-view image, and the coronary artery segmentation network comprises a first coronary artery segmentation model, a second coronary artery segmentation model and a third coronary artery segmentation model;
the segmentation module is configured to segment the clipped coronary CTA image by using the first coronary vessel segmentation model to obtain a first segmentation result, segment the clipped coronary CTA image by using the second coronary vessel segmentation model to obtain a second segmentation result, segment the clipped coronary CTA image by using the third coronary vessel segmentation model to obtain a third segmentation result, and merge the first segmentation result, the second segmentation result, and the third segmentation result to obtain a final coronary vessel segmentation result.
6. The apparatus of claim 5, wherein the cropping module crops the original CTA image according to the size of the heart, resulting in sub-images for coronary vessel segmentation, by:
the cropping module is specifically configured to retain a maximum connected domain of the heart segmentation result, and calculate an initial bounding box of the maximum connected domainThe coordinates of the initial bounding box are (x)0，x1，y0，y1，z0，z1)，x0，x1Denotes its coordinate range on the x-axis, y0，y1Denotes its coordinate range on the y-axis, z0，z1Representing its coordinate range on the z-axis; in the case where the original coronary CTA image is of size (W, H, D), the original bounding box is spread around by a physical distance, (W, H, D), and the coordinates of the new bounding box are determined: x'0＝max(0，x0-w/s0)；x′1＝min(W，x1+w/s0)；y′0＝max(0，y0-h/s1)；y′1＝min(H，y1+h/s1)；z′0＝max(0，z0-d/s2)；z′1＝min(D，z1+d/s2) (ii) a Wherein(s)0，s1，s2) Spacing information for the original coronary CTA image, the spacing information comprising: pixel pitch and layer pitch; and clipping the original coronary CTA image by using the new bounding box to obtain a clipped coronary CTA image.
7. The apparatus of claim 5, wherein the cropping module segments the heart by inputting an original coronary CTA image to obtain a heart segmentation result:
the clipping module is specifically configured to input the original coronary CTA image into a CNN network, segment a heart, and obtain the heart segmentation result.
8. The apparatus of claim 5, wherein the training module trains the coronary vessel segmentation network using the multi-view image by:
the training module is specifically used for taking a slice block along the axial position to train the first coronary vessel segmentation model; taking a slice block along a sagittal position to train the second coronary vessel segmentation model; and taking a slice block along the coronary position to train the third coronary vessel segmentation model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010855420.3A CN112037186A (en) | 2020-08-24 | 2020-08-24 | Coronary vessel extraction method and device based on multi-view model fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010855420.3A CN112037186A (en) | 2020-08-24 | 2020-08-24 | Coronary vessel extraction method and device based on multi-view model fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112037186A true CN112037186A (en) | 2020-12-04 |
Family
ID=73580673
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010855420.3A Pending CN112037186A (en) | 2020-08-24 | 2020-08-24 | Coronary vessel extraction method and device based on multi-view model fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112037186A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112561868A (en) * | 2020-12-09 | 2021-03-26 | 深圳大学 | Cerebrovascular segmentation method based on multi-view cascade deep learning network |
| CN112862835A (en) * | 2021-01-19 | 2021-05-28 | 杭州深睿博联科技有限公司 | Coronary vessel segmentation method, device, equipment and computer readable storage medium |
| CN115690309A (en) * | 2022-09-29 | 2023-02-03 | 中国人民解放军总医院第一医学中心 | Coronary artery CTA automatic three-dimensional post-processing method and device |
| WO2023071154A1 (en) * | 2021-10-29 | 2023-05-04 | 上海商汤智能科技有限公司 | Image segmentation method, training method and apparatus for related model, and device |
-
2020
- 2020-08-24 CN CN202010855420.3A patent/CN112037186A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112561868A (en) * | 2020-12-09 | 2021-03-26 | 深圳大学 | Cerebrovascular segmentation method based on multi-view cascade deep learning network |
| CN112561868B (en) * | 2020-12-09 | 2021-12-07 | 深圳大学 | Cerebrovascular segmentation method based on multi-view cascade deep learning network |
| CN112862835A (en) * | 2021-01-19 | 2021-05-28 | 杭州深睿博联科技有限公司 | Coronary vessel segmentation method, device, equipment and computer readable storage medium |
| WO2023071154A1 (en) * | 2021-10-29 | 2023-05-04 | 上海商汤智能科技有限公司 | Image segmentation method, training method and apparatus for related model, and device |
| CN115690309A (en) * | 2022-09-29 | 2023-02-03 | 中国人民解放军总医院第一医学中心 | Coronary artery CTA automatic three-dimensional post-processing method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112037186A (en) | Coronary vessel extraction method and device based on multi-view model fusion | |
| CN109345585B (en) | Method and system for measuring morphological parameters of intracranial aneurysm image | |
| US8483462B2 (en) | Object centric data reformation with application to rib visualization | |
| CN109472780B (en) | Method and system for measuring morphological parameters of intracranial aneurysm image | |
| Lu et al. | Timeline editing of objects in video | |
| CN111462071B (en) | Image processing method and system | |
| CN111028248A (en) | Method and device for separating static and dynamic pulses based on CT (computed tomography) image | |
| CN109472823B (en) | Method and system for measuring morphological parameters of intracranial aneurysm image | |
| CN111402150B (en) | CT image metal artifact removal method and device | |
| CN112991365B (en) | Coronary artery segmentation method, system and storage medium | |
| CN104783825A (en) | Method and apparatus for generating 2D projection image of vascular system | |
| US11769291B2 (en) | Method and device for rendering point cloud-based data | |
| EP2206093B1 (en) | Automatic movie fly-path calculation | |
| CN111915620B (en) | CT rib segmentation method and device | |
| Goobic et al. | Image stabilization and registration for tracking cells in the microvasculature | |
| CN114365190A (en) | Spleen tumor identification method based on VRDS 4D medical image and related device | |
| JP7333520B2 (en) | LEARNING PROGRAM, LEARNING METHOD, AND INFORMATION PROCESSING DEVICE | |
| CN109886973A (en) | A kind of vessel extraction method, apparatus and computer readable storage medium | |
| CN108655571A (en) | A kind of digital-control laser engraving machine, control system and control method, computer | |
| CN107871326B (en) | Image processing apparatus, image processing method, and storage medium | |
| CN111862001B (en) | Semi-automatic labeling method and device for CT images, electronic equipment and storage medium | |
| CN116433695B (en) | Mammary gland region extraction method and system of mammary gland molybdenum target image | |
| Vishal et al. | Unsupervised desmoking of laparoscopy images using multi-scale desmokenet | |
| CN109472803B (en) | Intracranial artery blood vessel segmentation method and system | |
| JP5473096B2 (en) | Modeling concave surfaces in image-based visual hulls |
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 |