CN112132850B - Vascular boundary detection method, system and device based on modal learning - Google Patents

Vascular boundary detection method, system and device based on modal learning Download PDF

Info

Publication number
CN112132850B
CN112132850B CN202010989178.9A CN202010989178A CN112132850B CN 112132850 B CN112132850 B CN 112132850B CN 202010989178 A CN202010989178 A CN 202010989178A CN 112132850 B CN112132850 B CN 112132850B
Authority
CN
China
Prior art keywords
image
model
modal
learning
detected
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
Application number
CN202010989178.9A
Other languages
Chinese (zh)
Other versions
CN112132850A (en
Inventor
高智凡
刘修健
张贺晔
徐梓峰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sun Yat Sen University
Original Assignee
Sun Yat Sen University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Sun Yat Sen University filed Critical Sun Yat Sen University
Priority to CN202010989178.9A priority Critical patent/CN112132850B/en
Publication of CN112132850A publication Critical patent/CN112132850A/en
Application granted granted Critical
Publication of CN112132850B publication Critical patent/CN112132850B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • 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
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • 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
    • G06T2207/10101—Optical tomography; Optical coherence tomography [OCT]
    • 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
    • G06T2207/10132—Ultrasound 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
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30101—Blood vessel; Artery; Vein; Vascular

Abstract

The application discloses a blood vessel boundary detection method, a system and a device based on modal learning, wherein the method comprises the following steps: acquiring batch images and dividing the images to obtain a hidden mode set and a target mode set; constructing a modal learning model based on the convolution automatic encoder and the bidirectional pyramid network structure; carrying out parameter optimization on the model learning model according to the hidden model set and the target model set to obtain an optimized model; and acquiring an image to be detected and inputting the image to be detected into the optimization model to identify the blood vessel boundary. The system comprises: the device comprises an acquisition module, a construction module, a modal learning module and an identification module. The apparatus comprises a memory and a processor for performing the above-described vessel boundary detection method based on modality learning. By using the application, accurate blood vessel boundary detection can be realized in various blood vessel environments. The method, the system and the device for detecting the blood vessel boundary based on modal learning can be widely applied to the field of blood vessel image processing.

Description

Vascular boundary detection method, system and device based on modal learning
Technical Field
The application relates to the field of blood vessel image processing, in particular to a blood vessel boundary detection method, system and device based on modal learning.
Background
Intracoronary imaging is an important medical imaging method in clinical practice, which presents high-resolution images of the internal morphology of the coronary arteries by inserting a miniature saturated imaging probe into the coronary arteries from outside the human body. It mainly includes two imaging modes: intravascular ultrasound (IVUS) and Optical Coherence Tomography (OCT) to show the manifestations of different aspects of coronary artery disease. Vessel Boundary Detection (VBDI) in intra-coronary images plays an important role in computer-aided image analysis, and it facilitates subsequent analysis, and existing vessel boundary detection methods based on hand-made image features are limited by vessel environment variability and cannot be well adapted to various vessel environments.
Disclosure of Invention
In order to solve the technical problems, the application aims to provide a blood vessel boundary detection method, a system and a device based on modal learning, which can be well adapted to different blood vessel environments, so that accurate blood vessel boundary detection can be still carried out in various blood vessel environments.
The first technical scheme adopted by the application is as follows: a blood vessel boundary detection method based on modal learning comprises the following steps:
acquiring batch images and dividing the images to obtain a hidden mode set and a target mode set;
constructing a modal learning model based on the convolution automatic encoder and the bidirectional pyramid network structure;
carrying out parameter optimization on the model learning model according to the hidden model set and the target model set to obtain an optimized model;
and acquiring an image to be detected, inputting the image to be detected into an optimization model, and identifying to obtain a blood vessel boundary.
Further, the modality learning model includes a first convolution automatic encoder for image segmentation of the lumen region, a second convolution automatic encoder for image segmentation of the lumen region and the media adventitia region, and a bi-directional pyramid network for image feature extraction.
Further, the step of optimizing parameters of the model learning model based on the hidden model set and the target model set to obtain an optimized model specifically includes:
inputting the hidden mode set and the target mode set into a mode learning model;
respectively extracting corresponding modal characteristics based on a modal learning model, carrying out characteristic fusion, and outputting processed target modal information;
optimizing soft marks and hard marks in model learning model parameters according to target modal information;
and repeating the optimizing step to obtain an optimizing model.
Further, the bi-directional pyramid network includes 5 concatenation modules, each comprising 4 inflated convolution layer structures.
Further, the step of acquiring an image to be measured and inputting the image to be measured into an optimization model to identify and obtain a blood vessel boundary specifically comprises the following steps:
acquiring an image to be detected and inputting the image to be detected into an optimization model;
performing regional image segmentation and image feature extraction on the image to be detected based on the optimization model, and outputting a target mode of the image to be detected;
and identifying and obtaining the blood vessel boundary according to the target mode of the image to be detected.
Further, the step of performing region image segmentation and image feature extraction on the image to be detected based on the optimization model and outputting the target mode of the image to be detected specifically further comprises the following steps:
image segmentation is carried out on the image to be detected based on a convolution encoder in the optimization model, so as to obtain a vascular lumen area;
image feature extraction is carried out on the vascular lumen area according to the soft mark and the hard mark in the model, and an extraction result is obtained;
and outputting a target mode of the image to be detected according to the extraction result.
The second technical scheme adopted by the application is as follows: a vessel boundary detection system based on modal learning, comprising the following modules:
the acquisition module is used for acquiring batch images and dividing the images into a hidden mode set and a target mode set;
the construction module is used for constructing a modal learning model based on the convolution automatic encoder and the bidirectional pyramid network structure;
the mode learning module is used for carrying out parameter optimization on the mode learning model according to the hidden mode set and the target mode set to obtain an optimized model;
the recognition module is used for acquiring the image to be detected and inputting the image to the optimization model, and recognizing and obtaining the blood vessel boundary.
The third technical scheme adopted by the application is as follows: a vessel boundary detection device based on modal learning, comprising:
at least one processor;
at least one memory for storing at least one program;
the at least one program, when executed by the at least one processor, causes the at least one processor to implement a method of vessel boundary detection based on modality learning as described above.
The method, the system and the device have the beneficial effects that: according to the application, parameters of the model learning model are optimized for multiple times through the hidden mode set and the target mode set until an optimized model conforming to a preset rule is obtained, and the accurate blood vessel boundary can be identified under the condition of single-mode input finally.
Drawings
FIG. 1 is a flow chart of steps of a method for detecting a blood vessel boundary based on modal learning;
fig. 2 is a block diagram of a blood vessel boundary detection system based on modal learning according to the present application.
Detailed Description
The application will now be described in further detail with reference to the drawings and to specific examples. The step numbers in the following embodiments are set for convenience of illustration only, and the order between the steps is not limited in any way, and the execution order of the steps in the embodiments may be adaptively adjusted according to the understanding of those skilled in the art.
As shown in fig. 1, the present application provides a blood vessel boundary detection method based on modal learning, which includes the following steps:
s1, acquiring batch images and dividing the images to obtain a hidden mode set and a target mode set;
s2, constructing a modal learning model based on a convolution automatic encoder and a bidirectional pyramid network structure;
s3, carrying out parameter optimization on the model learning model according to the hidden mode set and the target mode set to obtain an optimized model;
s4, acquiring an image to be detected, inputting the image to be detected into an optimization model, and identifying to obtain a blood vessel boundary.
Specifically, the method specifically comprises two stages, namely hidden mode learning and second mode distillation, wherein the hidden mode learning firstly inputs a hidden mode and a target mode in a model at the same time, extracts characteristic expressions in two imaging modes through the model, and further fuses the characteristics so as to output processed target mode information, and the stage converts the input and processing of one imaging mode into the input and processing of multiple modes, so that the problem of Single Input Single Task (SIST) is solved by a multi-input multi-task (MIMT) structure; the modal distillation is to identify the vessel boundary in the case where the knowledge learned by the model is applied to a single-mode input, i.e. in the case of only one image input.
Further as a preferred embodiment of the method, the modality learning model comprises a first convolution automatic encoder for image segmentation of lumen regions, a second convolution automatic encoder for image segmentation of lumen regions and media adventitia regions, and a bi-directional pyramid network for image feature extraction.
Specifically, in order to extract image features, identification of different areas of an image, that is, image segmentation, must be performed first. Only one region in the OCT image, namely the vessel lumen region, the IVUS image has two regions, namely the vessel lumen region and the media adventitia region (i.e. MA region), two convolutional automatic encoders are used in the process, which both fuse advanced image features, one is used for segmentation of the lumen region in the OCT image and the IVUS image, and the other is used for segmentation of the lumen and media adventitia region in the IVUS image, enhancing the adaptability of the model.
Additionally, a loss function is included to enrich the error types of network optimizations.
Further as a preferred embodiment of the method, the step of performing parameter optimization on the model learning model based on the hidden mode set and the target mode set to obtain an optimized model specifically includes:
inputting the hidden mode set and the target mode set into a mode learning model;
respectively extracting corresponding modal characteristics based on a modal learning model, carrying out characteristic fusion, and outputting processed target modal information;
optimizing soft marks and hard marks in model learning model parameters according to target modal information;
and repeating the optimizing step to obtain an optimizing model.
Specifically, given X t ，Y t ，{X pi ，Y pi }，i＝1，2，···，N；Y t Is subjected to X t Constraint of Y pi Is subjected to X pi Constraint, two groups of data have related learnable knowledge; the purpose is as follows: study how to get { X ] t ，X pi One-to-one correspondence of { Y } in the range Ω t ，Y pi And } on.
Wherein X is t ，Y t Training examples and marks of target modes respectively, N is the number of hidden modes, X pi ，Y pi Training examples and marks of the ith hidden mode are respectively, and omega is a parameter range of model learning.
From the above definition, hidden mode learning will be X t Corresponding to Y t Is converted into a dataset { X ] t ，X pi }。
Further as a preferred embodiment of the method, the bidirectional pyramid network comprises 5 cascade modules, each cascade module comprising 4 expansion convolution layer structures.
In particular, since the advanced feature maps of OCT and IVUS images have a fixed-size receive field, to accommodate this problem and maintain image resolution, the present application proposes a bi-directional pyramid network architecture. The structure is a dense connection structure of 5 cascade modules connected in the forward direction, each module containing 4 expansion convolution layers. The values of the expansion ratio in the block from low level to high level are respectively: 2. 2, 4 and 8. A pooling layer is added between every two adjacent modules for compressing the width and height of the image. After the image is subjected to expansion convolution, the receiving field can be enlarged, and the densely connected forward structure can reduce the loss of detailed information by combining low-level high-resolution features with high-level low-resolution features. In the reverse direction, the outputs of all forward blocks are connected from high to low to generate four feature maps. After convolution, each feature map is considered the output of the bi-directional pyramid network and then receives supervisory information (i.e., secondary supervision) from the markers during network training. The bidirectional pyramid network structure can effectively extract image information of blood vessel regions with variable sizes and preserve blood vessel boundaries.
Further as a preferred embodiment of the method, the step of obtaining the image to be measured and inputting the image to be measured into the optimization model to identify and obtain the blood vessel boundary specifically includes:
acquiring an image to be detected and inputting the image to be detected into an optimization model;
performing regional image segmentation and image feature extraction on the image to be detected based on the optimization model, and outputting a target mode of the image to be detected;
and identifying and obtaining the blood vessel boundary according to the target mode of the image to be detected.
Further as a preferred embodiment of the method, the step of performing region image segmentation and image feature extraction on the image to be detected based on the optimization model and outputting the target mode of the image to be detected specifically further includes:
image segmentation is carried out on the image to be detected based on a convolution encoder in the optimization model, so as to obtain a vascular lumen area;
image feature extraction is carried out on the vascular lumen area according to the soft mark and the hard mark in the model, and an extraction result is obtained;
and outputting a target mode of the image to be detected according to the extraction result.
As shown in fig. 2, a blood vessel boundary detection system based on modal learning includes:
the diagram structure module is used for generating a diagram structure according to the data relationship;
the acquisition module is used for acquiring batch images and dividing the images into a hidden mode set and a target mode set;
the construction module is used for constructing a modal learning model based on the convolution automatic encoder and the bidirectional pyramid network structure;
the mode learning module is used for carrying out parameter optimization on the mode learning model according to the hidden mode set and the target mode set to obtain an optimized model;
the recognition module is used for acquiring the image to be detected and inputting the image to the optimization model, and recognizing and obtaining the blood vessel boundary.
Further as a preferred embodiment of the system, the modality learning module further includes:
the input sub-module is used for inputting the hidden mode set and the target mode set into the mode learning model;
the modal processing sub-module is used for respectively extracting corresponding modal characteristics based on the modal learning model and carrying out characteristic fusion and outputting processed target modal information;
the optimizing sub-module is used for optimizing the soft mark and the hard mark in the model learning model parameters according to the target modal information;
and the cyclic sub-module is used for repeating the optimization steps to obtain an optimization model.
Further as a preferred embodiment of the system, the identification module further comprises:
the image acquisition submodule to be measured is used for acquiring an image to be measured and inputting the image to be measured into the optimization model;
the image processing sub-module to be detected is used for carrying out region image segmentation and image feature extraction on the image to be detected based on the optimization model and outputting a target mode of the image to be detected;
and the target mode sub-module of the image to be detected is used for identifying and obtaining the blood vessel boundary according to the target mode of the image to be detected.
The content in the method embodiment is applicable to the system embodiment, the functions specifically realized by the system embodiment are the same as those of the method embodiment, and the achieved beneficial effects are the same as those of the method embodiment.
A blood vessel boundary detection device based on modal learning:
at least one processor;
at least one memory for storing at least one program;
the at least one program, when executed by the at least one processor, causes the at least one processor to implement a method of vessel boundary detection based on modality learning as described above.
The content in the method embodiment is applicable to the embodiment of the device, and the functions specifically realized by the embodiment of the device are the same as those of the method embodiment, and the obtained beneficial effects are the same as those of the method embodiment.
While the preferred embodiment of the present application has been described in detail, the application is not limited to the embodiment, and various equivalent modifications and substitutions can be made by those skilled in the art without departing from the spirit of the application, and these equivalent modifications and substitutions are intended to be included in the scope of the present application as defined in the appended claims.

Claims (6)

1. A blood vessel boundary detection method based on modal learning is characterized by comprising the following steps:
acquiring batch images and dividing the images to obtain a hidden mode set and a target mode set;
constructing a modal learning model based on the convolution automatic encoder and the bidirectional pyramid network structure;
carrying out parameter optimization on the model learning model according to the hidden model set and the target model set to obtain an optimized model;
acquiring an image to be detected, inputting the image to be detected into an optimization model, and identifying to obtain a blood vessel boundary;
the modal learning model comprises a first convolution automatic encoder, a second convolution automatic encoder and a bidirectional pyramid network, wherein the first convolution automatic encoder is used for image segmentation of a lumen area, the second convolution automatic encoder is used for image segmentation of the lumen area and a medium membrane adventitia area, and the bidirectional pyramid network is used for image feature extraction;
the bidirectional pyramid network comprises 5 cascade modules, and each cascade module comprises 4 expansion convolution layer structures.
2. The method for detecting a blood vessel boundary based on modal learning according to claim 1, wherein the step of optimizing parameters of the model learning according to the hidden modal set and the target modal set to obtain an optimized model specifically comprises the steps of:
inputting the hidden mode set and the target mode set into a mode learning model;
respectively extracting corresponding modal characteristics based on a modal learning model, carrying out characteristic fusion, and outputting processed target modal information;
optimizing soft marks and hard marks in model learning model parameters according to target modal information;
and repeating the optimizing step to obtain an optimizing model.
3. The method for detecting a blood vessel boundary based on modal learning according to claim 2, wherein the step of acquiring the image to be detected and inputting the image to be detected into the optimization model to identify the blood vessel boundary comprises the following steps:
acquiring an image to be detected and inputting the image to be detected into an optimization model;
performing regional image segmentation and image feature extraction on the image to be detected based on the optimization model, and outputting a target mode of the image to be detected;
and identifying and obtaining the blood vessel boundary according to the target mode of the image to be detected.
4. The method for detecting a blood vessel boundary based on modal learning according to claim 3, wherein the step of performing region image segmentation and image feature extraction on the image to be detected based on the optimization model and outputting the target modality of the image to be detected specifically further comprises:
image segmentation is carried out on the image to be detected based on a convolution encoder in the optimization model, so as to obtain a vascular lumen area;
image feature extraction is carried out on the vascular lumen area according to the soft mark and the hard mark in the model, and an extraction result is obtained;
and outputting a target mode of the image to be detected according to the extraction result.
5. A vessel boundary detection system based on modal learning, comprising the following modules:
the acquisition module is used for acquiring batch images and dividing the images into a hidden mode set and a target mode set;
the construction module is used for constructing a modal learning model based on the convolution automatic encoder and the bidirectional pyramid network structure;
the mode learning module is used for carrying out parameter optimization on the mode learning model according to the hidden mode set and the target mode set to obtain an optimized model;
the recognition module is used for acquiring an image to be detected and inputting the image to the optimization model, and recognizing to obtain a blood vessel boundary;
the modal learning model comprises a first convolution automatic encoder, a second convolution automatic encoder and a bidirectional pyramid network, wherein the first convolution automatic encoder is used for image segmentation of a lumen area, the second convolution automatic encoder is used for image segmentation of the lumen area and a medium membrane adventitia area, and the bidirectional pyramid network is used for image feature extraction;
the bidirectional pyramid network comprises 5 cascade modules, and each cascade module comprises 4 expansion convolution layer structures.
6. A vessel boundary detection device based on modal learning, comprising:
at least one processor;
at least one memory for storing at least one program;
the at least one program, when executed by the at least one processor, causes the at least one processor to implement a method of vessel boundary detection based on modal learning as set forth in any one of claims 1-4.
CN202010989178.9A 2020-09-18 2020-09-18 Vascular boundary detection method, system and device based on modal learning Active CN112132850B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010989178.9A CN112132850B (en) 2020-09-18 2020-09-18 Vascular boundary detection method, system and device based on modal learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010989178.9A CN112132850B (en) 2020-09-18 2020-09-18 Vascular boundary detection method, system and device based on modal learning

Publications (2)

Publication Number Publication Date
CN112132850A CN112132850A (en) 2020-12-25
CN112132850B true CN112132850B (en) 2023-09-29

Family

ID=73842995

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010989178.9A Active CN112132850B (en) 2020-09-18 2020-09-18 Vascular boundary detection method, system and device based on modal learning

Country Status (1)

Country Link
CN (1) CN112132850B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20170113251A (en) * 2016-03-24 2017-10-12 재단법인 아산사회복지재단 Method and device for automatic inner and outer vessel wall segmentation in intravascular ultrasound images using deep learning
CN109859158A (en) * 2018-11-27 2019-06-07 邦鼓思电子科技（上海）有限公司 A kind of detection system, method and the machinery equipment on the working region boundary of view-based access control model

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10769791B2 (en) * 2017-10-13 2020-09-08 Beijing Keya Medical Technology Co., Ltd. Systems and methods for cross-modality image segmentation

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20170113251A (en) * 2016-03-24 2017-10-12 재단법인 아산사회복지재단 Method and device for automatic inner and outer vessel wall segmentation in intravascular ultrasound images using deep learning
CN109859158A (en) * 2018-11-27 2019-06-07 邦鼓思电子科技（上海）有限公司 A kind of detection system, method and the machinery equipment on the working region boundary of view-based access control model

Also Published As

Publication number Publication date
CN112132850A (en) 2020-12-25

Similar Documents

Publication Publication Date Title
CN110517238B (en) AI three-dimensional reconstruction and human-computer interaction visualization network system for CT medical image
CN111627017B (en) Automatic segmentation method for vascular lumen based on deep learning
CN111612756B (en) Coronary artery specificity calcification detection method and device
CN111784762B (en) Method and device for extracting blood vessel center line of X-ray radiography image
CN111009032B (en) Vascular three-dimensional reconstruction method based on improved epipolar line constraint matching
CN112037146B (en) Automatic correction method and device for medical image artifacts and computer equipment
CN110310280A (en) Hepatic duct and the image-recognizing method of calculus, system, equipment and storage medium
CN111192320B (en) Position information determining method, device, equipment and storage medium
CN111667459A (en) Medical sign detection method, system, terminal and storage medium based on 3D variable convolution and time sequence feature fusion
CN111178420B (en) Coronary artery segment marking method and system on two-dimensional contrast image
CN112132834A (en) Ventricular image segmentation method, system, device and storage medium
CN116228792A (en) Medical image segmentation method, system and electronic device
JP7349018B2 (en) Coronary artery segmentation method, apparatus, electronic device and computer readable storage medium
CN111242969A (en) Boundary node determination method, grid division method and medical equipment
CN112132850B (en) Vascular boundary detection method, system and device based on modal learning
CN110599444A (en) Device, system and non-transitory readable storage medium for predicting fractional flow reserve of a vessel tree
CN113936138A (en) Target detection method, system, equipment and medium based on multi-source image fusion
CN111599448A (en) Multi-view shape constraint system and method for specific coronary artery calcification analysis
CN111402207A (en) Ultrasonic angiography video data analysis method based on composite neural network
CN115409859A (en) Coronary artery blood vessel image segmentation method and device, storage medium and terminal
CN116071270A (en) Electronic data generation method and system for generating countermeasure network based on deformable convolution
CN114240846A (en) System and method for reducing false positive rate of medical image focus segmentation result
CN113379770B (en) Construction method of nasopharyngeal carcinoma MR image segmentation network, image segmentation method and device
CN114998582A (en) Coronary artery blood vessel segmentation method, device and storage medium
CN114052762A (en) Method for predicting size of narrow blood vessel and size of instrument based on Swin-T

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