CN110827255A - Plaque stability prediction method and system based on coronary artery CT image - Google Patents
Plaque stability prediction method and system based on coronary artery CT image Download PDFInfo
- Publication number
- CN110827255A CN110827255A CN201911053961.8A CN201911053961A CN110827255A CN 110827255 A CN110827255 A CN 110827255A CN 201911053961 A CN201911053961 A CN 201911053961A CN 110827255 A CN110827255 A CN 110827255A
- Authority
- CN
- China
- Prior art keywords
- plaque
- image
- neural network
- coronary artery
- different resolutions
- 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
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
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus for radiation diagnosis, e.g. combined with radiation therapy equipment
- A61B6/02—Devices for diagnosis sequentially in different planes; Stereoscopic radiation diagnosis
- A61B6/03—Computerised tomographs
- A61B6/032—Transmission computed tomography [CT]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus for radiation diagnosis, e.g. combined with radiation therapy equipment
- A61B6/50—Clinical applications
- A61B6/504—Clinical applications involving diagnosis of blood vessels, e.g. by angiography
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus for radiation diagnosis, e.g. combined with radiation therapy equipment
- A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
- A61B6/5205—Devices using data or image processing specially adapted for radiation diagnosis involving processing of raw data to produce diagnostic data
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus for radiation diagnosis, e.g. combined with radiation therapy equipment
- A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
- A61B6/5211—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data
- A61B6/5217—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data extracting a diagnostic or physiological parameter from medical diagnostic data
-
- 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
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30101—Blood vessel; Artery; Vein; Vascular
Abstract
The invention discloses a plaque stability prediction method and a plaque stability prediction system based on a coronary artery CT image, wherein the plaque stability prediction method comprises the following steps: determining a plaque location based on a coronary CT image of the target object; extracting a plaque and a surrounding fat image based on the plaque position; acquiring image blocks with different resolutions based on the plaque and the surrounding fat image; and obtaining a plaque stability prediction result through a neural network model based on image blocks with different resolutions. The plaque stability prediction method and system based on the coronary artery CT image determine the position of the plaque based on the coronary artery CT image, extract the plaque and the peripheral fat image, further obtain the plaque stability prediction result through the neural network model, do not need an ultrasonic probe to measure the plaque property in the blood vessel, do not damage the blood vessel of a patient, have high automation degree, are efficient and convenient, can greatly reduce the pain and economic burden of the patient, and have great clinical application potential.
Description
Technical Field
The invention belongs to the technical field of artificial intelligence, and particularly relates to a plaque stability prediction method and system based on a coronary artery CT image.
Background
Unstable coronary plaque refers to a portion of the coronary artery with mild to moderate stenosis and a tendency to rupture or ulcerate, which can often cause coronary heart disease, myocardial infarction, sudden cardiac death, and the like. Therefore, the search and prevention of unstable coronary plaque are important means for preventing and treating coronary heart disease, and the prediction of plaque stability is crucial.
At present, the property of the plaque is measured by mainly utilizing an intravascular ultrasonic probe through a catheter operation clinically, and the attribute of the plaque is further judged. Therefore, it is particularly desirable to develop a convenient, efficient, highly automated method for noninvasive plaque prediction.
Disclosure of Invention
The invention aims to provide a plaque stability prediction method and a plaque stability prediction system based on a coronary artery CT image, which can be used for conveniently, efficiently and highly automatically predicting plaques without wound.
According to an aspect of the present invention, a plaque stability prediction method based on a coronary artery CT image is provided, including: determining a plaque location based on a coronary CT image of the target object; extracting a plaque and a surrounding fat image based on the plaque position; acquiring image blocks with different resolutions based on the plaque and the surrounding fat image; and obtaining a plaque stability prediction result through a neural network model based on the image blocks with different resolutions.
Preferably, the acquiring image blocks with different resolutions based on the plaque and the surrounding fat image comprises: and performing sampling processing on the extracted plaque and the surrounding fat image for multiple times to obtain image blocks with different resolutions.
Preferably, the neural network model is obtained by: acquiring a plurality of existing coronary artery CT images and intravascular ultrasound examination results corresponding to each existing coronary artery CT image; determining a plaque position aiming at each existing coronary artery CT image, extracting plaque and surrounding fat images based on the plaque position, and acquiring image blocks with different resolutions corresponding to the plaque and the surrounding fat images; and (3) taking image blocks with different resolutions corresponding to each existing coronary artery CT image as model input, taking the corresponding intravascular ultrasound examination result as a golden standard of a model output result, and training by using a random gradient descent method to obtain the neural network model.
Preferably, the neural network model comprises a plurality of data inputs and a feature extraction sub-network corresponding to each data input.
Preferably, the image blocks with different resolutions are respectively input into different data input ends of the neural network model, and corresponding image features are obtained through corresponding feature extraction sub-networks; and acquiring a final plaque stability prediction result by utilizing a fully-connected network layer of the neural network and a Softmax function based on image features corresponding to the image blocks with different resolutions.
Preferably, the plaque and the surrounding fat image in the cube with the preset side length are extracted by taking the plaque position as the center.
According to another aspect of the present invention, a plaque stability prediction system based on a coronary CT image is provided, the system comprising: a memory storing computer-executable instructions; a processor executing computer executable instructions in the memory to perform the steps of: determining a plaque location based on a coronary CT image of the target object; extracting a plaque and a surrounding fat image based on the plaque position; acquiring image blocks with different resolutions based on the plaque and the surrounding fat image; and obtaining a plaque stability prediction result through a neural network model based on the image blocks with different resolutions.
Preferably, the acquiring image blocks with different resolutions based on the plaque and the surrounding fat image comprises: and performing sampling processing on the extracted plaque and the surrounding fat image for multiple times to obtain image blocks with different resolutions.
Preferably, the neural network model is obtained by: acquiring a plurality of existing coronary artery CT images and intravascular ultrasound examination results corresponding to each existing coronary artery CT image; determining a plaque position aiming at each existing coronary artery CT image, extracting plaque and surrounding fat images based on the plaque position, and acquiring image blocks with different resolutions corresponding to the plaque and the surrounding fat images; and (3) taking image blocks with different resolutions corresponding to each existing coronary artery CT image as model input, taking the corresponding intravascular ultrasound examination result as a gold standard of a model output result, and training by using a random gradient descent method to obtain the neural network model.
Preferably, the neural network model comprises a plurality of data inputs and a feature extraction sub-network corresponding to each data input.
Preferably, the image blocks with different resolutions are respectively input into different data input ends of the neural network model, and corresponding image features are obtained through corresponding feature extraction sub-networks; and acquiring a final plaque stability prediction result by utilizing a fully-connected network layer of the neural network and a Softmax function based on image features corresponding to the image blocks with different resolutions.
Preferably, the plaque and the surrounding fat image in the cube with the preset side length are extracted by taking the plaque position as the center.
The invention has the beneficial effects that: the plaque stability prediction method and system based on the coronary artery CT image determine the position of the plaque based on the coronary artery CT image, extract the plaque and the peripheral fat image, further obtain the plaque stability prediction result through the neural network model, do not need an ultrasonic probe to measure the plaque property in the blood vessel, do not damage the blood vessel of a patient, have high automation degree, are efficient and convenient, can greatly reduce the pain and economic burden of the patient, and have great clinical application potential.
The method and system of the present invention have other features and advantages which will be apparent from or are set forth in detail in the accompanying drawings and the following detailed description, which are incorporated herein, and which together serve to explain certain principles of the invention.
Drawings
The above and other objects, features and advantages of the present invention will become more apparent by describing in more detail exemplary embodiments thereof with reference to the attached drawings, in which like reference numerals generally represent like parts throughout.
Fig. 1 shows a flowchart of a plaque stability prediction method based on a coronary CT image according to an embodiment of the present invention.
Detailed Description
Preferred embodiments of the present invention will be described in more detail below. While the following describes preferred embodiments of the present invention, it should be understood that the present invention may be embodied in various forms and should not be limited by 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 invention to those skilled in the art.
The plaque stability prediction method based on the coronary artery CT image comprises the following steps: determining a plaque location based on a coronary CT image of the target object; extracting a plaque and a surrounding fat image based on the plaque position; acquiring image blocks with different resolutions based on the plaque and the surrounding fat image; and obtaining a plaque stability prediction result through a neural network model based on image blocks with different resolutions.
Pericoronary fat (PCAT) refers to the adipose tissue surrounding the adventitia of the three major branches of the coronary artery. The PCAT generates active substances such as fat factors, inflammatory factors and the like through a paracrine way, regulates the reconstruction of vascular walls and the inflammation around the blood vessels, participates in the plaque formation and evolution process, and is expected to become a new target point for coronary heart disease diagnosis and treatment. Research has proved that PCAT phenotypic characteristics are different among coronary atherosclerotic plaques with different properties, and the method has great significance for evaluating the vulnerability of the plaques.
The renaissance of neural networks starting from AlexNet has been continued for over ten years, during which time deep neural networks have achieved remarkable results in the fields of computer vision, natural language processing, games, etc., and with the affirmation of the charming prize, the academic position of deep learning has finally come to true fate. In a plurality of fields of deep learning, medical image analysis can mean a wind-borne disease, and the deep learning has unusual expression in various aspects such as pulmonary nodule screening, breast cancer pathological picture identification, skin disease picture classification, diabetic retinopathy detection, Alzheimer's disease diagnosis and the like. Although deep learning has advanced a long way in the field of medical images, no work has been involved in the plaque prediction problem.
Specifically, the plaque position is determined through manual identification based on a coronary artery CT image of a target object, plaque and peripheral fat images are extracted based on the plaque position, image blocks with different resolutions are obtained based on the plaque and the peripheral fat images and serve as data materials for subsequently analyzing the plaque stability, and then a plaque stability prediction result is obtained through a neural network model.
According to an exemplary embodiment, the plaque stability prediction method based on the coronary artery CT image selects a target plaque position based on the coronary artery CT imaging, extracts plaque and peripheral fat image data, uses the data as a data material for subsequently analyzing the plaque stability, analyzes the plaque and the peripheral fat image by using the deep neural network to obtain a plaque stability prediction result, does not need an ultrasonic probe to measure the plaque property in a blood vessel, does not damage the blood vessel of a patient, has high automation degree, is efficient and convenient, can greatly reduce the pain and economic burden of the patient, and has considerable clinical application potential.
Preferably, the acquiring image blocks with different resolutions based on the plaque and the surrounding fat image comprises: and performing sampling processing on the extracted plaque and the surrounding fat image for multiple times to obtain image blocks with different resolutions.
Specifically, the extracted patch and the surrounding fat image are respectively subjected to sampling processing for multiple times, and the image blocks subjected to sampling processing each time have different resolutions, so that a plurality of image blocks with different resolutions are obtained.
Preferably, the neural network model is obtained by the following steps: acquiring a plurality of existing coronary artery CT images and intravascular ultrasound examination results corresponding to each existing coronary artery CT image; determining the position of a plaque aiming at each existing coronary artery CT image, extracting the plaque and a surrounding fat image based on the position of the plaque, and acquiring image blocks with different resolutions corresponding to the plaque and the surrounding fat image; and (3) taking image blocks with different resolutions corresponding to each existing coronary artery CT image as model input, taking the corresponding ultrasonic examination result as a golden standard of a model output result, and training by using a random gradient descent method to obtain a neural network model.
Deep neural networks extract simple features from the original input and on this basis progressively extract more complex abstract features until the final result. This process of extracting abstract features of an object may need to go through many times, this is called depth.
Specifically, based on the existing coronary artery CT images and the corresponding intravascular ultrasound examination results, the plaque position is determined for each existing coronary artery CT image, the plaque and the peripheral fat images are extracted based on the plaque position, different-resolution image blocks corresponding to the plaque and the peripheral fat images are further acquired, the different-resolution image blocks corresponding to each existing coronary artery CT image are used as data input of the neural network model, the corresponding intravascular ultrasound examination results are used as a golden standard of the output results of the neural network model, and the random gradient descent method is used for training, so that the neural network model is further acquired.
The neural network is a multi-path convolution neural network, cross entropy is used as a loss function of model training, and a random gradient descent algorithm is used for training the model.
Preferably, the neural network model comprises a plurality of data inputs and a feature extraction sub-network corresponding to each data input.
Specifically, the neural network model is a multi-channel neural network model, and comprises a plurality of data input ends and a feature extraction sub-network corresponding to each data input end, and image features corresponding to the data input ends are obtained through the corresponding feature extraction networks.
As a preferred scheme, the image blocks with different resolutions are respectively input into different data input ends of a neural network model, and corresponding image features are obtained through corresponding feature extraction sub-networks; and acquiring a final plaque stability prediction result by utilizing a fully-connected network layer of the neural network and a Softmax function based on the image characteristics corresponding to the image blocks with different resolutions.
Specifically, the image blocks with different resolutions are used as different data input ends of the neural network model, image features corresponding to the data input ends are obtained through corresponding feature extraction sub-networks, and plaque stability prediction results are obtained by using the full connection layer and the Softmax function based on the image features corresponding to the different data input ends. Each sub-network is formed by combining modules such as a convolution layer, a pooling layer and an up-sampling layer, and when the model is trained through a random gradient descent algorithm, the network can automatically learn the image characteristics with the most judgment capability.
Preferably, the plaque and fat images in a cube with a preset side length are extracted by taking the plaque position as the center.
Specifically, the plaque and fat images in a cube with a preset side length are extracted by taking the plaque position as a center.
In one example, the serological biochemical parameters, the physical sign parameters are input as data inputs of the neural network model.
Specifically, parameters such as patient serum biochemical substances, physical signs and the like are input into the deep neural network, and plaque stability is jointly predicted by combining the images.
In one example, plaque stability prediction results are included in the diagnostic report.
According to another aspect of the present invention, a plaque stability prediction system based on a coronary CT image is provided, the system comprising: a memory storing computer-executable instructions; a processor executing computer executable instructions in the memory to perform the steps of: determining a plaque location based on a coronary CT image of the target object; extracting a plaque and a surrounding fat image based on the plaque position; acquiring image blocks with different resolutions based on the plaque and the surrounding fat image; and obtaining a plaque stability prediction result through a neural network model based on image blocks with different resolutions.
Pericoronary fat (PCAT) refers to the adipose tissue surrounding the adventitia of the three major branches of the coronary artery. The PCAT generates active substances such as fat factors, inflammatory factors and the like through a paracrine way, regulates the reconstruction of vascular walls and the inflammation around the blood vessels, participates in the plaque formation and evolution process, and is expected to become a new target point for coronary heart disease diagnosis and treatment. Research has proved that PCAT phenotypic characteristics are different among coronary atherosclerotic plaques with different properties, and the method has great significance for evaluating the vulnerability of the plaques.
The renaissance of neural networks starting from AlexNet has been continued for over ten years, during which time deep neural networks have achieved remarkable results in the fields of computer vision, natural language processing, games, etc., and with the affirmation of the charming prize, the academic position of deep learning has finally come to true fate. In a plurality of fields of deep learning, medical image analysis can mean a wind-borne disease, and the deep learning has unusual expression in various aspects such as pulmonary nodule screening, breast cancer pathological picture identification, skin disease picture classification, diabetic retinopathy detection, Alzheimer's disease diagnosis and the like. Although deep learning has advanced a long way in the field of medical images, no work has been involved in the plaque prediction problem.
Specifically, the plaque position is determined through manual identification based on a coronary artery CT image of a target object, the plaque position can also be determined through intelligent image identification, the plaque and the peripheral fat image are extracted based on the plaque position, image blocks with different resolutions are obtained based on the plaque and the peripheral fat image and are used as data materials for subsequently analyzing the plaque stability, and then a plaque stability prediction result is obtained through a neural network model.
According to an exemplary embodiment, the plaque stability prediction method based on the coronary artery CT image selects a target plaque position based on the coronary artery CT imaging, extracts plaque and peripheral fat image data, uses the data as a data material for subsequently analyzing the plaque stability, analyzes the plaque and the peripheral fat image by using the deep neural network to obtain a plaque stability prediction result, does not need an ultrasonic probe to measure the plaque property in a blood vessel, does not damage the blood vessel of a patient, has high automation degree, is efficient and convenient, can greatly reduce the pain and economic burden of the patient, and has considerable clinical application potential.
Preferably, the acquiring image blocks with different resolutions based on the plaque and the surrounding fat image comprises: and performing sampling processing on the extracted plaque and the surrounding fat image for multiple times to obtain image blocks with different resolutions.
Specifically, the extracted patch and the surrounding fat image are respectively subjected to sampling processing for multiple times, and the image blocks subjected to sampling processing each time have different resolutions, so that a plurality of image blocks with different resolutions are obtained.
Preferably, the neural network model is obtained by the following steps: acquiring a plurality of existing coronary artery CT images and intravascular ultrasound examination results corresponding to each existing coronary artery CT image; determining the position of a plaque aiming at each existing coronary artery CT image, extracting the plaque and a surrounding fat image based on the position of the plaque, and acquiring image blocks with different resolutions corresponding to the plaque and the surrounding fat image; and (3) taking image blocks with different resolutions corresponding to each existing coronary artery CT image as model input, taking the corresponding ultrasonic examination result as a golden standard of a model output result, and training by using a random gradient descent method to obtain a neural network model.
Deep neural networks extract simple features from the original input and on this basis progressively extract more complex abstract features until the final result. This process of extracting abstract features of an object may need to go through many times, this is called depth.
Specifically, based on the existing coronary artery CT images and the corresponding intravascular ultrasound examination results, the plaque position is determined for each existing coronary artery CT image, the plaque and the peripheral fat images are extracted based on the plaque position, different-resolution image blocks corresponding to the plaque and the peripheral fat images are further acquired, the different-resolution image blocks corresponding to each existing coronary artery CT image are used as data input of the neural network model, the corresponding intravascular ultrasound examination results are used as a golden standard of the output results of the neural network model, and the random gradient descent method is used for training, so that the neural network model is further acquired.
The neural network is a multi-path convolution neural network, cross entropy is used as a loss function of model training, and a random gradient descent algorithm is used for training the model.
Preferably, the neural network model comprises a plurality of data inputs and a feature extraction sub-network corresponding to each data input.
Specifically, the neural network model is a multi-channel neural network model, and comprises a plurality of data input ends and a feature extraction sub-network corresponding to each data input end, and image features corresponding to the data input ends are obtained through the corresponding feature extraction networks.
As a preferred scheme, the image blocks with different resolutions are respectively input into different data input ends of a neural network model, and corresponding image features are obtained through corresponding feature extraction sub-networks; and acquiring a final plaque stability prediction result by utilizing a fully-connected network layer of the neural network and a Softmax function based on the image characteristics corresponding to the image blocks with different resolutions.
Specifically, the image blocks with different resolutions are used as different data input ends of the neural network model, image features corresponding to the data input ends are obtained through corresponding feature extraction sub-networks, and plaque stability prediction results are obtained by using the full connection layer and the Softmax function based on the image features corresponding to the different data input ends. Each sub-network is formed by combining modules such as a convolution layer, a pooling layer and an up-sampling layer, and when the model is trained through a random gradient descent algorithm, the network can automatically learn the image characteristics with the most judgment capability.
Preferably, the plaque and fat images in the cube with the preset side length are extracted by taking the plaque position as the center.
Specifically, the plaque and fat images in a preset side length cube are extracted by taking the plaque position as the center.
In one example, the serological biochemical parameters, the physical sign parameters are input as data inputs of the neural network model.
Specifically, parameters such as patient serum biochemical substances, physical signs and the like are input into the deep neural network, and plaque stability is jointly predicted by combining the images.
In one example, plaque stability prediction results are included in the diagnostic report.
In one example, the prediction system is installed on a post-processing workstation or other computer that can be interconnected with the CT device or the PACS system.
In particular, the prediction system communicates with a CT or PACS system from which coronary CT data is captured and displayed in a list format. An operator only needs to select a case to be analyzed in the system, double-click opens the image, click the position of a target plaque in the image, the system can automatically extract a corresponding image area, and the result is analyzed and calculated to give a plaque stability prediction result.
Examples
Fig. 1 shows a flowchart of a plaque stability prediction method based on a coronary CT image according to an embodiment of the present invention.
As shown in fig. 1, the plaque stability prediction method based on the coronary CT image includes:
s102: determining a plaque location based on a coronary CT image of the target object;
s104: extracting a plaque and a surrounding fat image based on the plaque position;
extracting plaque and fat images in a preset side length cube by taking the plaque position as a center;
s106: acquiring image blocks with different resolutions based on the plaque and the surrounding fat image;
wherein, obtaining image blocks with different resolutions based on the plaque and the surrounding fat image comprises: sampling the extracted plaque and surrounding fat images for multiple times to obtain image blocks with different resolutions;
s108: obtaining a plaque stability prediction result through a neural network model based on image blocks with different resolutions;
wherein the neural network model is obtained by the steps of: acquiring a plurality of existing coronary artery CT images and intravascular ultrasound examination results corresponding to each existing coronary artery CT image; determining the position of a plaque aiming at each existing coronary artery CT image, extracting the plaque and a surrounding fat image based on the position of the plaque, and acquiring image blocks with different resolutions corresponding to the plaque and the surrounding fat image; and (3) taking image blocks with different resolutions corresponding to each existing coronary artery CT image as model input, taking the corresponding intravascular ultrasound examination result as a golden standard of the output result of the neural network model, and training by using a random gradient descent method to obtain the neural network model.
The neural network model comprises a plurality of data input ends and a feature extraction sub-network corresponding to each data input end;
respectively inputting image blocks with different resolutions into different data input ends of a neural network model, and acquiring corresponding image characteristics through corresponding characteristic extraction sub-networks; and acquiring a final plaque stability prediction result by utilizing a fully-connected network layer of the neural network and a Softmax function based on the image characteristics corresponding to the image blocks with different resolutions.
Having described embodiments of the present invention, the foregoing description is intended to be exemplary, not exhaustive, and not limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments.
Claims (10)
1. A plaque stability prediction method based on a coronary artery CT image is characterized by comprising the following steps:
determining a plaque location based on a coronary CT image of the target object;
extracting a plaque and a surrounding fat image based on the plaque position;
acquiring image blocks with different resolutions based on the plaque and the surrounding fat image;
and obtaining a plaque stability prediction result through a neural network model based on the image blocks with different resolutions.
2. The plaque stability prediction method of claim 1 wherein the obtaining different resolution patches based on the plaque and surrounding fat images comprises:
and performing sampling processing on the extracted plaque and the surrounding fat image for multiple times to obtain image blocks with different resolutions.
3. The plaque stability prediction method of claim 1 wherein the neural network model is obtained by:
acquiring a plurality of existing coronary artery CT images and intravascular ultrasound examination results corresponding to each existing coronary artery CT image;
determining a plaque position aiming at each existing coronary artery CT image, extracting plaque and surrounding fat images based on the plaque position, and acquiring image blocks with different resolutions corresponding to the plaque and the surrounding fat images;
and (3) taking image blocks with different resolutions corresponding to each existing coronary artery CT image as model input, taking the corresponding intravascular ultrasound examination result as a golden standard of a model output result, and training by using a random gradient descent method to obtain the neural network model.
4. The plaque stability prediction method of claim 2 wherein the neural network model comprises a plurality of data inputs and a feature extraction sub-network corresponding to each data input.
5. The plaque stability prediction method of claim 4 wherein image patches of different resolutions are respectively input to different data input ends of the neural network model, and corresponding image features are obtained through corresponding feature extraction sub-networks; and acquiring a final plaque stability prediction result by utilizing a fully-connected network layer of the neural network and a Softmax function based on image features corresponding to the image blocks with different resolutions.
6. The plaque-stability predicting method according to claim 1, wherein the plaque and the peripheral fat image in a cube of a predetermined side length are extracted with the plaque position as a center.
7. A system for predicting plaque stability based on CT images of coronary arteries, the system comprising:
a memory storing computer-executable instructions;
a processor executing computer executable instructions in the memory to perform the steps of:
determining a plaque location based on a coronary CT image of the target object;
extracting a plaque and a surrounding fat image based on the plaque position;
acquiring image blocks with different resolutions based on the plaque and the surrounding fat image;
and obtaining a plaque stability prediction result through a neural network model based on the image blocks with different resolutions.
8. The plaque stability prediction system of claim 7 wherein the obtaining different resolution patches based on the plaque and surrounding fat images comprises:
and performing sampling processing on the extracted plaque and the surrounding fat image for multiple times to obtain image blocks with different resolutions.
9. The plaque stability prediction system of claim 8 wherein the neural network model is obtained by:
acquiring a plurality of existing coronary artery CT images and intravascular ultrasound examination results corresponding to each existing coronary artery CT image;
determining a plaque position aiming at each existing coronary artery CT image, extracting plaque and surrounding fat images based on the plaque position, and acquiring image blocks with different resolutions corresponding to the plaque and the surrounding fat images;
and (3) taking image blocks with different resolutions corresponding to each existing coronary artery CT image as model input, taking the corresponding intravascular ultrasound examination result as a golden standard of a model output result, and training by using a random gradient descent method to obtain the neural network model.
10. The plaque stability prediction system of claim 8 wherein fat images within a cube of a predetermined side length are extracted centered on the plaque location.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911053961.8A CN110827255A (en) | 2019-10-31 | 2019-10-31 | Plaque stability prediction method and system based on coronary artery CT image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911053961.8A CN110827255A (en) | 2019-10-31 | 2019-10-31 | Plaque stability prediction method and system based on coronary artery CT image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110827255A true CN110827255A (en) | 2020-02-21 |
Family
ID=69551822
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911053961.8A Pending CN110827255A (en) | 2019-10-31 | 2019-10-31 | Plaque stability prediction method and system based on coronary artery CT image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110827255A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113077432A (en) * | 2021-03-30 | 2021-07-06 | 中国人民解放军空军军医大学 | Patient risk grading system based on coronary artery CTA image atherosclerotic plaque comprehensive characteristics |
| CN113657479A (en) * | 2021-08-12 | 2021-11-16 | 广东省人民医院 | Novel multi-scale depth-width combined pathological picture classification method, system and medium |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130094749A1 (en) * | 2011-10-14 | 2013-04-18 | Fujifilm Corporation | Model-based coronary artery calcium scoring |
| CN103917166A (en) * | 2011-08-17 | 2014-07-09 | Vp诊断公司 | A method and system of characterization of carotid plaque |
| CN106056126A (en) * | 2015-02-13 | 2016-10-26 | 西门子公司 | Plaque vulnerability assessment in medical imaging |
| CN106974622A (en) * | 2017-04-06 | 2017-07-25 | 上海交通大学 | plaque stability measuring method and system based on optical coherence tomography |
| CN107730497A (en) * | 2017-10-27 | 2018-02-23 | 哈尔滨工业大学 | A kind of plaque within blood vessels property analysis method based on depth migration study |
| CN107993228A (en) * | 2017-12-15 | 2018-05-04 | 中国人民解放军总医院 | A kind of vulnerable plaque automatic testing method and device based on cardiovascular OCT images |
| CN108257674A (en) * | 2018-01-24 | 2018-07-06 | 龙马智芯（珠海横琴）科技有限公司 | Disease forecasting method and apparatus, equipment, computer readable storage medium |
| CN108268870A (en) * | 2018-01-29 | 2018-07-10 | 重庆理工大学 | Multi-scale feature fusion ultrasonoscopy semantic segmentation method based on confrontation study |
| CN110223781A (en) * | 2019-06-03 | 2019-09-10 | 中国医科大学附属第一医院 | A kind of various dimensions plaque rupture Warning System |
-
2019
- 2019-10-31 CN CN201911053961.8A patent/CN110827255A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103917166A (en) * | 2011-08-17 | 2014-07-09 | Vp诊断公司 | A method and system of characterization of carotid plaque |
| US20130094749A1 (en) * | 2011-10-14 | 2013-04-18 | Fujifilm Corporation | Model-based coronary artery calcium scoring |
| CN106056126A (en) * | 2015-02-13 | 2016-10-26 | 西门子公司 | Plaque vulnerability assessment in medical imaging |
| CN106974622A (en) * | 2017-04-06 | 2017-07-25 | 上海交通大学 | plaque stability measuring method and system based on optical coherence tomography |
| CN107730497A (en) * | 2017-10-27 | 2018-02-23 | 哈尔滨工业大学 | A kind of plaque within blood vessels property analysis method based on depth migration study |
| CN107993228A (en) * | 2017-12-15 | 2018-05-04 | 中国人民解放军总医院 | A kind of vulnerable plaque automatic testing method and device based on cardiovascular OCT images |
| CN108257674A (en) * | 2018-01-24 | 2018-07-06 | 龙马智芯（珠海横琴）科技有限公司 | Disease forecasting method and apparatus, equipment, computer readable storage medium |
| CN108268870A (en) * | 2018-01-29 | 2018-07-10 | 重庆理工大学 | Multi-scale feature fusion ultrasonoscopy semantic segmentation method based on confrontation study |
| CN110223781A (en) * | 2019-06-03 | 2019-09-10 | 中国医科大学附属第一医院 | A kind of various dimensions plaque rupture Warning System |
Non-Patent Citations (2)
| Title |
|---|
| NICOLA GAIBAZZI 等: "Coronary inflammation by computed tomography pericoronary fat attenuation in MINOCA and tako -tsubo syndrome", JAHA * |
| 杨本强 等: "冠状动脉周围脂肪：冠心病临床诊疗"新靶点"", 中国介入影像与治疗学, vol. 16, no. 4, pages 234 - 236 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113077432A (en) * | 2021-03-30 | 2021-07-06 | 中国人民解放军空军军医大学 | Patient risk grading system based on coronary artery CTA image atherosclerotic plaque comprehensive characteristics |
| CN113077432B (en) * | 2021-03-30 | 2024-01-05 | 中国人民解放军空军军医大学 | Patient risk grading system based on coronary artery CTA image atherosclerosis plaque comprehensive characteristics |
| CN113657479A (en) * | 2021-08-12 | 2021-11-16 | 广东省人民医院 | Novel multi-scale depth-width combined pathological picture classification method, system and medium |
| CN113657479B (en) * | 2021-08-12 | 2022-12-06 | 广东省人民医院 | Novel multi-scale depth-width combined pathological picture classification method, system and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11576621B2 (en) | Plaque vulnerability assessment in medical imaging | |
| US11101033B2 (en) | Medical image aided diagnosis method and system combining image recognition and report editing | |
| Tan et al. | Application of stacked convolutional and long short-term memory network for accurate identification of CAD ECG signals | |
| Sharif et al. | A comprehensive review on multi-organs tumor detection based on machine learning | |
| KR102212499B1 (en) | Ivus image analysis method | |
| JP5039294B2 (en) | Apparatus and method for analyzing tissue type along a cylindrical structure | |
| TWI668666B (en) | Prediction model for grouping hepatocellular carcinoma, prediction system thereof, and method for determining hepatocellular carcinoma group | |
| US20150003706A1 (en) | Probability mapping for visualisation and analysis of biomedical images | |
| JP2016531709A (en) | Image analysis technology for diagnosing disease | |
| WO2010117576A2 (en) | Image processing and machine learning for diagnostic analysis of microcirculation | |
| CN111178369B (en) | Medical image recognition method and system, electronic equipment and storage medium | |
| CN110827255A (en) | Plaque stability prediction method and system based on coronary artery CT image | |
| JPWO2020027228A1 (en) | Diagnostic support system and diagnostic support method | |
| CN110353639B (en) | Blood supply area quantification method and system based on blood vessel enhanced radiography | |
| CN113470060A (en) | Coronary artery multi-angle curved surface reconstruction visualization method based on CT image | |
| KR20230118054A (en) | Method and system for predicting biomarker expression in medical image | |
| JP2019513449A (en) | Method of obtaining 3D retinal blood vessel morphology from optical coherence tomography images and method of analyzing them | |
| Dehnavi et al. | Automatic soft and hard plaque detection in IVUS images: A textural approach | |
| Hernanda et al. | Semantic segmentation of venous on deep vein thrombosis (DVT) case using UNet-ResNet | |
| CN116309346A (en) | Medical image detection method, device, equipment, storage medium and program product | |
| JP2004283583A (en) | Operation method of image forming medical inspection system | |
| JPH09299366A (en) | Region extract device | |
| Salehi et al. | Investigation and simulation of different medical image processing algorithms to improve image quality using simulink matlab | |
| Mi et al. | Detecting carotid intima-media from small-sample ultrasound images | |
| Chacko et al. | Sequential functional analysis of left ventricle from 2D-echocardiography images |
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 |