CN114298971A - Coronary artery segmentation method, system, terminal and storage medium - Google Patents
Coronary artery segmentation method, system, terminal and storage medium Download PDFInfo
- Publication number
- CN114298971A CN114298971A CN202111394930.6A CN202111394930A CN114298971A CN 114298971 A CN114298971 A CN 114298971A CN 202111394930 A CN202111394930 A CN 202111394930A CN 114298971 A CN114298971 A CN 114298971A
- Authority
- CN
- China
- Prior art keywords
- coronary artery
- segmentation
- dimensional
- image
- skeleton
- 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
Abstract
The present application relates to a coronary artery segmentation method, system, terminal and storage medium. The method comprises the following steps: performing initial segmentation on a coronary artery image by using a two-dimensional convolution network to obtain a rough coronary artery segmentation result, and performing three-dimensional reconstruction on the rough coronary artery segmentation result; performing skeleton extraction on the three-dimensional reconstructed image by using a skeleton thinning algorithm, and intercepting an interested region on the coronary artery image by taking an extracted skeleton point as a center; and segmenting the region of interest by utilizing a three-dimensional convolution network, and performing multi-point growth on the segmentation result to obtain a coronary artery segmentation result. According to the method and the device, the segmentation is focused on the blood vessel region through skeleton extraction, the interference of non-blood vessels is reduced, the coronary artery is automatically segmented, the regions which cannot be identified by other networks can be identified, the phenomena of blood vessel disconnection and blood vessel deletion are avoided, and a better segmentation effect is obtained.
Description
Technical Field
The present application relates to the field of medical image processing technologies, and in particular, to a coronary artery segmentation method, system, terminal, and storage medium.
Background
The enhancement and segmentation of blood vessels is a long-standing task in medical image analysis, and conventional coronary artery segmentation methods include region growing methods, active contour models, statistical models, shape models, particle filtering, path tracking, and the like. These conventional coronary segmentation methods are all interactive and require the provision of seed points. Coronary artery segmentation methods without interaction include level set, graph cut, and the like. However, since coronary segmentation is considered a voxel-based classification problem, the above method may eventually yield a large number of false positives or false negatives.
Since 2016, the application of deep learning in medical images has been greatly developed, and various deep learning segmentation networks, such as FCN (full Convolution Network), UNet, 3D UNet or Res UNet, have been proposed, and have obtained better segmentation results than the conventional methods in the segmentation field. However, when the coronary artery segmentation is performed by using the deep learning segmentation network, since the coronary artery is composed of an extremely fine coronary structure and the feature of the segmented region is not uniform due to the stenosis of the blood vessel, the segmented region is easily disconnected from the blood vessel and is easily lost.
Disclosure of Invention
The present application provides a coronary artery segmentation method, system, terminal and storage medium, which are intended to solve at least one of the above technical problems in the prior art to some extent.
In order to solve the above problems, the present application provides the following technical solutions:
a coronary artery segmentation method comprising:
performing initial segmentation on a coronary artery image by using a two-dimensional convolution network to obtain a rough coronary artery segmentation result, and performing three-dimensional reconstruction on the rough coronary artery segmentation result;
performing skeleton extraction on the three-dimensional reconstructed image by using a skeleton thinning algorithm, and intercepting an interested region on the coronary artery image by taking an extracted skeleton point as a center;
and segmenting the region of interest by utilizing a three-dimensional convolution network, and performing multi-point growth on the segmentation result to obtain a coronary artery segmentation result.
The technical scheme adopted by the embodiment of the application further comprises the following steps: the coronary images are CTA images.
The technical scheme adopted by the embodiment of the application further comprises the following steps: the two-dimensional convolutional network takes Deeplab v3+ as a basic network, the Deeplab v3+ comprises a cavity space convolutional pooling pyramid, and the cavity space convolutional pooling pyramid comprises a cavity convolutional layer, a convolutional layer and a global average pooling layer.
The technical scheme adopted by the embodiment of the application further comprises the following steps: the training process of the two-dimensional convolutional network specifically comprises the following steps:
the input of the two-dimensional convolution network is an original cross section of a CTA image, the original cross section is used as a central cross section, a set number of connected original cross sections are taken as channels of the two-dimensional convolution network for supplementary input, and the central cross section is used as a label for training.
The technical scheme adopted by the embodiment of the application further comprises the following steps: the skeleton extraction of the three-dimensional reconstructed image by using a skeleton thinning algorithm specifically comprises the following steps:
listing a candidate pixel list to be removed;
scanning the image, and deleting pixels in the candidate pixel list;
and rechecking the pixels in the candidate pixel list by taking the connectivity of the maintained image as an index, and iteratively scanning the image until the image stops changing.
The technical scheme adopted by the embodiment of the application further comprises the following steps: the method for segmenting the region of interest by utilizing the three-dimensional convolution network and performing multi-seed point growth on the segmentation result specifically comprises the following steps:
inputting the region of interest into a three-dimensional convolution network for segmentation, and performing three-dimensional reconstruction on the segmentation result of the three-dimensional convolution network;
removing a small connected region of the three-dimensional reconstructed image according to the extracted skeleton;
and taking the skeleton point without the small connected region as an initial seed point of multiple sub-points, performing multiple sub-point growth on the segmentation result of the three-dimensional convolution network, and performing three-dimensional space mapping according to the multiple sub-point growth result to obtain a coronary artery segmentation result.
The technical scheme adopted by the embodiment of the application further comprises the following steps: the specific steps of carrying out multi-seed point growth on the segmentation result of the three-dimensional convolution network are as follows:
taking the multiple seed points as a set, and adding the segmentation result of the 26 neighborhoods of the seed points into the set of the blood vessels;
and repeatedly adding the segmentation results of the 26 neighborhoods to the points of the blood vessels by using the updated set until no blood vessel point is added, thereby obtaining the segmentation result of the coronary artery.
Another technical scheme adopted by the embodiment of the application is as follows: a coronary artery segmentation system comprising:
a two-dimensional segmentation module: the system comprises a two-dimensional convolution network, a rough coronary artery segmentation module, a three-dimensional reconstruction module and a data processing module, wherein the two-dimensional convolution network is used for carrying out initial segmentation on a coronary artery image, obtaining a rough coronary artery segmentation result and carrying out three-dimensional reconstruction on the rough coronary artery segmentation result;
a skeleton extraction module: the image processing device is used for extracting the skeleton of the three-dimensional reconstructed image by using a skeleton thinning algorithm and intercepting an interested region on the coronary artery image by taking the extracted skeleton point as the center;
a three-dimensional segmentation module: and the method is used for segmenting the region of interest by utilizing a three-dimensional convolution network and performing multi-point growth on the segmentation result to obtain a coronary artery segmentation result.
The embodiment of the application adopts another technical scheme that: a terminal comprising a processor, a memory coupled to the processor, wherein,
the memory stores program instructions for implementing the coronary artery segmentation method;
the processor is for executing the program instructions stored by the memory to control coronary artery segmentation.
The embodiment of the application adopts another technical scheme that: a storage medium storing program instructions executable by a processor to perform the coronary artery segmentation method.
Compared with the prior art, the embodiment of the application has the advantages that: the method, the system, the terminal and the storage medium for coronary artery segmentation in the embodiment of the application use a method of combining a two-dimensional convolution network and a three-dimensional convolution neural network to perform coronary artery segmentation, firstly, the two-dimensional convolution network is used for performing rough segmentation, and a segmentation result is used for extracting a skeleton, so that segmentation is focused on a blood vessel region, interference of a non-blood vessel is reduced, and then, the three-dimensional convolution network is used for performing coronary artery segmentation by taking a skeleton point as a center. The coronary artery segmentation method and the coronary artery segmentation device have the advantages that the coronary artery is automatically segmented, the regions which cannot be identified by other networks can be identified, the phenomena of blood vessel disconnection and blood vessel loss are avoided, and accordingly a better segmentation effect is achieved.
Drawings
FIG. 1 is a flow chart of a coronary artery segmentation method according to an embodiment of the present application;
FIG. 2 is a schematic diagram of a coronary artery segmentation network according to an embodiment of the present application;
FIG. 3 is a graph showing the variation of Loss and DSC (Dice Similarity Coefficient) with training turns of 3D Unet in the training set and the testing set according to the embodiment of the present application;
FIG. 4 is a schematic diagram showing the segmentation effect of the tag, 3D Unet, Res Unet and the segmentation result of the embodiment on the CTA cross section
FIG. 5 is a schematic diagram of a 3D reconstruction result of a tag, a 3D Unet, a Res Unet, and a segmentation result according to an embodiment of the present application;
FIG. 6 is a schematic structural diagram of a coronary artery segmentation system according to an embodiment of the present application;
fig. 7 is a schematic structural diagram of a terminal according to an embodiment of the present application;
fig. 8 is a schematic structural diagram of a storage medium according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
Aiming at the defects of the prior art, the coronary artery segmentation method provided by the embodiment of the application carries out coronary artery segmentation by using a method of combining a two-dimensional convolution network and a three-dimensional convolution neural network, firstly, a rough segmentation result is obtained by combining the two-dimensional convolution network, then, skeleton extraction is carried out on the rough segmentation result, and finally, a region is divided by taking a skeleton point as a center and the coronary artery segmentation is carried out by using the three-dimensional convolution network, so that the interference of a non-blood vessel region is reduced, and more blood vessel regions can be directly extracted.
Specifically, please refer to fig. 1, which is a flowchart illustrating a coronary artery segmentation method according to an embodiment of the present application. The coronary artery segmentation method comprises the following steps:
s1: acquiring a coronary artery image;
in this step, the coronary artery image is a CTA (CT angiography) image.
S2: performing primary segmentation on the coronary artery image by using a two-dimensional convolution network to obtain a rough coronary artery segmentation result, and performing three-dimensional reconstruction on the rough coronary artery segmentation result;
in this step, the input of the two-dimensional convolutional network is the original cross section of the CTA image, and the data size of the input image is [ 512512 ]. Fig. 2 is a schematic diagram of a coronary artery segmentation network implemented in the present application. In order to utilize two-dimensional plane information as much as possible, the two-dimensional convolutional network in the embodiment of the present application uses Deeplab v3+ with a larger network depth and a larger backhaul frame as a backbone network as a base network, and meanwhile, the Deeplab v3+ further includes an ASPP (aperture Spatial convolutional Pooling Pyramid) module optimized and improved for multi-scale information, where the improved ASPP module includes a 3 × 3 hole convolutional layer, a 1 × 1 convolutional layer, and a global average Pooling layer, and the hole rates of the hole convolutional layers are 12, 24, and 38, respectively. 5 output results are obtained through 3 void convolutional layers, 1 convolutional layer of 1 multiplied by 1 and a global average pooling layer, the 5 output results are connected, and finally the number of channels is reduced through one convolutional layer of 1 multiplied by 1.
And performing bilinear interpolation on the feature map obtained by the ASPP module, performing 3 x 3 convolution calculation, splicing the calculated result with the result of the first block of the resnet, performing 3 x 3 convolution calculation, finally performing bilinear interpolation on the calculated result to form the shape of the input picture, and performing convolution to obtain a prediction result. Assuming that the interpolation point coordinates (x, y) are known, the value f (x, y) of the interpolation point needs to be found. Knowing the 4 coordinate points (x) where the insertion point is closest1,y1),(x2,y2),(x3,y3),(x4,y4) The value of the interpolation point can be calculated by the following formula:
in the embodiment of the present application, the purpose of the two-dimensional convolutional network is to identify the coronary artery as accurately as possible, so that the weighted cross entropy loss is used as the loss function of the two-dimensional convolutional network, and the weighted ratio is the ratio of the background to the foreground. In order to make up for the information loss of the two-dimensional convolutional network in the three-dimensional space, the original cross section of a CTA image is taken as a central cross section, a set number of connected CT cross sections are taken as channels of the two-dimensional convolutional network for supplementary input, and the central cross section is taken as a label for model training. Since too much supplementary information may have too large a difference from the predicted region to hinder the segmentation, the number of the original cross sections taken as the channel supplement in the embodiment of the present application is three. The sensitivity of the original Deeplab v3+ network is 0.725, the sensitivity is 0.937 after the improvement of the embodiment of the application, and the identified blood vessels become more.
In this step, the three-dimensional reconstruction method specifically includes: and combining the segmentation results of the CT cross section of each frame by frame to obtain a three-dimensional reconstructed image.
S3: performing skeleton extraction on the three-dimensional reconstructed image by using a skeleton thinning algorithm, and intercepting a region with a set size on the CTA image by taking a skeleton point as a center to serve as an interested region;
in this step, the size of the region of interest is set to [32,64, 64 ]. The blood vessel in each region of interest occupies 38.3% of the whole region of interest on average in the largest cube, 87.98% in the largest occupation ratio and 0% in the smallest occupation ratio, so that each region of interest not only includes certain background information to facilitate identification, but also does not have the phenomenon that part of the blood vessel is not segmented. The skeleton extraction process comprises the following steps: the method comprises the steps of firstly listing a candidate pixel list to be removed, scanning an image, deleting pixels in the candidate pixel list, then rechecking the pixels in the candidate pixel list by taking the continuity of the image as an index, and iterating and scanning the image until the image stops changing. According to the embodiment of the application, the segmentation is focused on the blood vessel region through skeleton extraction, so that the interference of a non-blood vessel region is reduced, and the identification of the blood vessel region which is difficult to identify is facilitated.
S4: inputting the region of interest into a three-dimensional convolution network for segmentation, performing three-dimensional reconstruction on a segmentation result, and removing a small connected region of a reconstructed image according to the extracted skeleton;
in the network training stage, a skeleton label of an input image is extracted by using a skeleton thinning algorithm, skeleton points are extracted by taking 1 as an interval in a skeleton point set, a block with the size of [ 326464 ] is cut out from the input image by taking the skeleton points as a cutting center, and the cut block is added into a segmentation region to perform coronary artery segmentation. In the embodiment of the application, the loss function Dice of the three-dimensional convolution network is a loss function.
Since enlarging the size of the cropping block can increase the effective perception of feature points far from the cropping center, and using a cropping block of size [ 326464 ] has an effect on the prediction effect, the size of the cropping block is set to [ 62128128 ] in the model test stage. Specifically, as shown in fig. 3, it is a curve of variation of Loss and DSC (Dice Similarity Coefficient) along with training rounds of 3D Unet in the embodiment of the present application on the training set and the test set.
In the embodiment of the present application, the removal standard of the small connected region is as follows: and in the framework region, 26 neighborhoods are used as a connected mode, and the connected framework pixel points are less than 280 framework objects.
S5: taking the skeleton point without the small connected region as an initial seed point of the multiple sub-points, performing multiple sub-point growth on the segmentation result of the three-dimensional convolution network, and performing three-dimensional space mapping according to the multiple sub-point growth result to obtain a final coronary artery segmentation result;
in this step, the growing method of the multiple seed points is as follows: and taking the various seed points as a set, adding the 26-neighborhood segmentation results of the seed points into the blood vessel set, and repeatedly adding the 26-neighborhood segmentation results into the updated set as the blood vessel points until no blood vessel point is added. The final set is the final coronary artery segmentation result.
To further verify the feasibility and effectiveness of the embodiments of the present application, experiments were performed in the following examples, and to evaluate the performance of the method, the experimental model evaluation indexes include Dice Similarity Coefficient (DSC), sensitivity, Area under the ROC Curve (Area under the Curve of ROC, AUC), and Hausdorff Distance (HD), and the calculation formula of each evaluation index is as follows:
HD(A，B)＝max(h(A，B)，h(B，A) (4)
wherein TP, FP, FN are the number of voxels of true positive (labeled as foreground predicted as background), false positive (labeled as background predicted as foreground) and false negative (labeled as foreground predicted as background), respectively; HD describes a measure of the degree of similarity between two sets of points and is a defining form of the distance between the two sets of points. Assume that there are two point sets A, B, h (A, B) and h (B, A) describing the one-way Hausdorff distance from the A point set to the B point set and from the B point set to the A point set, respectively. | | | represents the distance paradigm from point set a to point set B, the euclidean distance used herein. The maximum value of the minimum distance from each point set to another point set described in HD is generally indicated by HD95 to prevent extreme cases from happening, which means that the maximum value of the minimum distance is not selected, but the minimum distance of the 0.95 th position after sorting by the minimum distance is selected as the index.
As shown in fig. 4, which is a schematic diagram of the segmentation effect of the label, the 3D Unet, the Res Unet and the segmentation result of the embodiment of the present application on the CTA cross section, each row is a CT cross section, (a) is the segmentation effect of the label, (b) is the segmentation effect after the 3D Unet post-treatment removes the impurities, (c) is the segmentation effect after the Res Unet post-treatment removes the impurities, and (D) is the segmentation effect of the embodiment of the present application.
Fig. 5 is a schematic diagram showing a label, 3D Unet, Res Unet, and a 3D reconstruction result of a segmentation result of an embodiment of the present application, where (a) is a three-dimensional reconstruction result of the label, (b) is a three-dimensional reconstruction result of a segmentation result after the 3D Unet is post-processed to remove impurities, (c) is a three-dimensional reconstruction result of a segmentation result after the Res Unet is post-processed to remove impurities, and (D) is a three-dimensional reconstruction result using a segmentation result of an embodiment of the present application. The arrows point to more identified coronary arteries in the present embodiment relative to the 3D Unet, Res Unet. The experimental result shows that compared with the 3D Unet and Res Unet, the coronary artery which can not be identified by other networks can be identified by the embodiment of the application, and the problems of disconnection and vessel loss are solved to a certain extent.
Based on the above, in the coronary artery segmentation method in the embodiment of the present application, a method in which a two-dimensional convolution network and a three-dimensional convolution neural network are combined is used to perform coronary artery segmentation, the two-dimensional convolution network is used to perform rough segmentation, and a segmentation result is used to extract a skeleton, so that segmentation is focused on a blood vessel region, thereby reducing non-blood vessel interference, and then the three-dimensional convolution network is used to perform coronary artery segmentation with a skeleton point as a center. The coronary artery segmentation method and the coronary artery segmentation device have the advantages that the coronary artery is automatically segmented, the regions which cannot be identified by other networks can be identified, the phenomena of blood vessel disconnection and blood vessel loss are avoided, and a better segmentation effect is achieved.
Please refer to fig. 6, which is a schematic structural diagram of a coronary artery segmentation system according to an embodiment of the present application. The coronary artery segmentation system 40 of the embodiment of the present application includes:
the two-dimensional segmentation module 41: the system comprises a two-dimensional convolution network, a rough coronary artery segmentation module, a three-dimensional reconstruction module and a data processing module, wherein the two-dimensional convolution network is used for carrying out initial segmentation on a coronary artery image, obtaining a rough coronary artery segmentation result and carrying out three-dimensional reconstruction on the rough coronary artery segmentation result;
the skeleton extraction module 42: the image processing device is used for extracting the skeleton of the three-dimensional reconstructed image by using a skeleton thinning algorithm and intercepting an interested region on the coronary artery image by taking the extracted skeleton point as the center;
the three-dimensional segmentation module 43: and the method is used for segmenting the region of interest by utilizing a three-dimensional convolution network and performing multi-point growth on the segmentation result to obtain a coronary artery segmentation result.
Please refer to fig. 7, which is a schematic diagram of a terminal structure according to an embodiment of the present application. The terminal 50 comprises a processor 51, a memory 52 coupled to the processor 51.
The memory 52 stores program instructions for implementing the coronary artery segmentation method described above.
The processor 51 is operative to execute program instructions stored in the memory 52 to control coronary artery segmentation.
The processor 51 may also be referred to as a CPU (Central Processing Unit). The processor 51 may be an integrated circuit chip having signal processing capabilities. The processor 51 may also be a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
Please refer to fig. 8, which is a schematic structural diagram of a storage medium according to an embodiment of the present application. The storage medium of the embodiment of the present application stores a program file 61 capable of implementing all the methods described above, where the program file 61 may be stored in the storage medium in the form of a software product, and includes several instructions to enable a computer device (which may be a personal computer, a server, or a network device) or a processor (processor) to execute all or part of the steps of the methods of the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a mobile hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, or terminal devices, such as a computer, a server, a mobile phone, and a tablet.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A coronary artery segmentation method, comprising:
performing initial segmentation on a coronary artery image by using a two-dimensional convolution network to obtain a rough coronary artery segmentation result, and performing three-dimensional reconstruction on the rough coronary artery segmentation result;
performing skeleton extraction on the three-dimensional reconstructed image by using a skeleton thinning algorithm, and intercepting an interested region on the coronary artery image by taking an extracted skeleton point as a center;
and segmenting the region of interest by utilizing a three-dimensional convolution network, and performing multi-point growth on the segmentation result to obtain a coronary artery segmentation result.
2. The coronary artery segmentation method according to claim 1, wherein the coronary artery image is a CTA image.
3. The coronary artery segmentation method of claim 2 wherein the two-dimensional convolutional network is based on deep v3+, the deep v3+ comprises a void space convolutional pooling pyramid comprising a void convolutional layer, a convolutional layer, and a global average pooling layer.
4. The coronary artery segmentation method according to claim 3, wherein the training process of the two-dimensional convolutional network is specifically:
the input of the two-dimensional convolution network is an original cross section of a CTA image, the original cross section is used as a central cross section, a set number of connected original cross sections are taken as channels of the two-dimensional convolution network for supplementary input, and the central cross section is used as a label for training.
5. The coronary artery segmentation method according to any one of claims 1 to 4, wherein the skeleton extraction of the three-dimensional reconstructed image by using a skeleton refinement algorithm is specifically:
listing a candidate pixel list to be removed;
scanning the image, and deleting pixels in the candidate pixel list;
and rechecking the pixels in the candidate pixel list by taking the connectivity of the maintained image as an index, and iteratively scanning the image until the image stops changing.
6. The coronary artery segmentation method according to claim 5, wherein the segmenting the region of interest by using the three-dimensional convolution network and performing the multi-seed point growing on the segmentation result specifically comprises:
inputting the region of interest into a three-dimensional convolution network for segmentation, and performing three-dimensional reconstruction on the segmentation result of the three-dimensional convolution network;
removing a small connected region of the three-dimensional reconstructed image according to the extracted skeleton;
and taking the skeleton point without the small connected region as an initial seed point of multiple sub-points, performing multiple sub-point growth on the segmentation result of the three-dimensional convolution network, and performing three-dimensional space mapping according to the multiple sub-point growth result to obtain a coronary artery segmentation result.
7. The coronary artery segmentation method according to claim 6, wherein the performing of the multi-seed growth on the segmentation result of the three-dimensional convolution network is specifically:
taking the multiple seed points as a set, and adding the segmentation result of the 26 neighborhoods of the seed points into the set of the blood vessels;
and repeatedly adding the segmentation results of the 26 neighborhoods to the points of the blood vessels by using the updated set until no blood vessel point is added, thereby obtaining the segmentation result of the coronary artery.
8. A coronary artery segmentation system, comprising:
a two-dimensional segmentation module: the system comprises a two-dimensional convolution network, a rough coronary artery segmentation module, a three-dimensional reconstruction module and a data processing module, wherein the two-dimensional convolution network is used for carrying out initial segmentation on a coronary artery image, obtaining a rough coronary artery segmentation result and carrying out three-dimensional reconstruction on the rough coronary artery segmentation result;
a skeleton extraction module: the image processing device is used for extracting the skeleton of the three-dimensional reconstructed image by using a skeleton thinning algorithm and intercepting an interested region on the coronary artery image by taking the extracted skeleton point as the center;
a three-dimensional segmentation module: and the method is used for segmenting the region of interest by utilizing a three-dimensional convolution network and performing multi-point growth on the segmentation result to obtain a coronary artery segmentation result.
9. A terminal, comprising a processor, a memory coupled to the processor, wherein,
the memory stores program instructions for implementing the coronary artery segmentation method of any one of claims 1-7;
the processor is for executing the program instructions stored by the memory to control coronary artery segmentation.
10. A storage medium having stored thereon program instructions executable by a processor to perform the coronary artery segmentation method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111394930.6A CN114298971A (en) | 2021-11-23 | 2021-11-23 | Coronary artery segmentation method, system, terminal and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111394930.6A CN114298971A (en) | 2021-11-23 | 2021-11-23 | Coronary artery segmentation method, system, terminal and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114298971A true CN114298971A (en) | 2022-04-08 |
Family
ID=80966334
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111394930.6A Pending CN114298971A (en) | 2021-11-23 | 2021-11-23 | Coronary artery segmentation method, system, terminal and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114298971A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115810015A (en) * | 2023-02-09 | 2023-03-17 | 慧影医疗科技(北京)股份有限公司 | Automatic knee joint segmentation method, system, medium and equipment based on deep learning |
| WO2023232137A1 (en) * | 2022-06-02 | 2023-12-07 | 北京阅影科技有限公司 | Method and apparatus for training image processing model, and method and apparatus for image processing |
-
2021
- 2021-11-23 CN CN202111394930.6A patent/CN114298971A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023232137A1 (en) * | 2022-06-02 | 2023-12-07 | 北京阅影科技有限公司 | Method and apparatus for training image processing model, and method and apparatus for image processing |
| CN115810015A (en) * | 2023-02-09 | 2023-03-17 | 慧影医疗科技(北京)股份有限公司 | Automatic knee joint segmentation method, system, medium and equipment based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111445478B (en) | Automatic intracranial aneurysm region detection system and detection method for CTA image | |
| CN110033410B (en) | Image reconstruction model training method, image super-resolution reconstruction method and device | |
| CN109410219B (en) | Image segmentation method and device based on pyramid fusion learning and computer readable storage medium | |
| CN110889852B (en) | Liver segmentation method based on residual error-attention deep neural network | |
| CN110599500B (en) | Tumor region segmentation method and system of liver CT image based on cascaded full convolution network | |
| US20110007933A1 (en) | 3D Image Processing | |
| CN114298971A (en) | Coronary artery segmentation method, system, terminal and storage medium | |
| CN110866938B (en) | Full-automatic video moving object segmentation method | |
| CN111369567B (en) | Method and device for segmenting target object in three-dimensional image and electronic equipment | |
| CN113436211A (en) | Medical image active contour segmentation method based on deep learning | |
| WO2021114870A1 (en) | Parallax estimation system and method, electronic device and computer-readable storage medium | |
| CN113421240B (en) | Mammary gland classification method and device based on ultrasonic automatic mammary gland full-volume imaging | |
| CN115965750B (en) | Vascular reconstruction method, vascular reconstruction device, vascular reconstruction computer device, and vascular reconstruction program | |
| CN110570394A (en) | medical image segmentation method, device, equipment and storage medium | |
| CN113159236A (en) | Multi-focus image fusion method and device based on multi-scale transformation | |
| Chen et al. | Single depth image super-resolution using convolutional neural networks | |
| Pan et al. | Multi-stage feature pyramid stereo network-based disparity estimation approach for two to three-dimensional video conversion | |
| Jian et al. | Dual-Branch-UNet: A Dual-Branch Convolutional Neural Network for Medical Image Segmentation. | |
| CN113313728B (en) | Intracranial artery segmentation method and system | |
| CN114494230A (en) | Breast focus segmentation device, model training method and electronic equipment | |
| CN110827963A (en) | Semantic segmentation method for pathological image and electronic equipment | |
| CN113570573A (en) | Pulmonary nodule false positive eliminating method, system and equipment based on mixed attention mechanism | |
| CN113052849A (en) | Automatic segmentation method and system for abdominal tissue image | |
| CN114565631A (en) | Plant leaf fine segmentation method based on double-layer convolution network and mask refinement | |
| CN113409324A (en) | Brain segmentation method fusing differential geometric information |
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 |