A cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition

Abstract

The present invention discloses a cardiovascular early warning method based on vulnerable plaque image recognition, comprising the following steps: acquiring cardiovascular and cerebrovascular data of a patient; performing image preprocessing and data annotation on the cardiovascular and cerebrovascular data of the patient to acquire three-dimensional cardiovascular and cerebrovascular data, and establishing a vulnerable plaque recognition model based on the three-dimensional cardiovascular and cerebrovascular data using a first convolutional neural network; extracting three-dimensional vascular images, establishing a fluid simulation model for the three-dimensional vascular images, and acquiring blood flow information using the fluid simulation model; acquiring rupture causes and early warning information based on the output of the vulnerable plaque recognition model and the blood flow information, and inputting the rupture causes and the early warning information into a second convolutional neural network model to construct an early warning model; the method is conducive to directly and conveniently acquiring detailed data information of vulnerable plaques, saving manpower and material resources for diagnosis and treatment; and providing an accurate and powerful reference for doctors to diagnose and treat vulnerable plaques, and is conducive to early detection of high-risk populations for cardiovascular and cerebrovascular diseases.

Description

Technical Field

[0001] The present invention relates to the technical field of medical engineering, and in particular to a cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition. Background Art

[0002] In recent years, with the continuous upgrading of medical technology and the continuous updating of computer technology, the combination of computer technology and medical technology has become closer, solving many medical problems. At the same time, massive medical data also provides strong data support for the prevention and treatment of diseases. Vulnerable plaque is an important pathological basis for acute coronary syndrome, and its rupture is an important cause of death. Therefore, accurate identification of vulnerable plaques based on medical data is an important measure to prevent cardiovascular and cerebrovascular diseases.

[0003] At present, a Chinese invention with publication number CN117198514B provides a vulnerable plaque recognition method and system based on the CLIP model. On the basis of the CLIP model, a BN layer and a Dropout layer are introduced to process text features and image features respectively, reducing overfitting. Vulnerable plaques are identified only based on image feature extraction. The recognition results and recognition criteria are too single, which is not conducive to accurately understanding the details of vulnerable plaques. There is no corresponding warning for the recognition results of vulnerable plaques, which is not conducive to the precise prevention and treatment of diseases. Summary of the invention

[0004] The technical problem solved by the present invention is that the existing technology only identifies vulnerable plaques based on image feature extraction, and the identification results and identification standards are too simple, which is not conducive to accurately understanding the details of the vulnerable plaques, and there is no corresponding warning for the identification results of the vulnerable plaques, which is not conducive to the accurate prevention and treatment of the disease.

[0005] In order to solve the above technical problems, the present invention provides a cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition, comprising the following steps:

[0006] Step S1, using big data technology to obtain the patient's cardiovascular and cerebrovascular data from the medical sharing platform, and establishing a vascular imaging database based on the patient's cardiovascular and cerebrovascular data;

[0007] Step S2, performing image preprocessing and data annotation on the patient's cardiovascular data to obtain three-dimensional cardiovascular data, and establishing a vulnerable plaque recognition model using a first convolutional neural network based on the three-dimensional cardiovascular data;

[0008] Step S3, extracting a three-dimensional blood vessel image from the three-dimensional cardiovascular and cerebrovascular data, establishing a three-dimensional fluid simulation model for the three-dimensional blood vessel image, and obtaining blood flow information using the three-dimensional fluid simulation model, wherein the blood flow information includes resting blood flow information, congestive blood flow information, and a blood flow reserve coefficient;

[0009] Step S4, obtaining rupture causes and warning information according to the output of the vulnerable plaque recognition model and the blood flow information, and inputting the rupture causes and the warning information into a second convolutional neural network model to construct a warning model;

[0010] As a preferred solution of the cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition described in the present invention, the step S1 specifically includes the following steps:

[0011] Step S101, using big data technology to obtain cardiovascular and cerebrovascular data of patients in a high-risk group from a medical sharing platform according to disease tags, wherein the high-risk group includes patients with high-risk factors for coronary heart disease, patients with high-risk factors for cardiovascular and cerebrovascular atherosclerosis, and patients with acute myocardial infarction, wherein the patients with high-risk factors for coronary heart disease include patients with diabetes, patients with hyperlipidemia, patients with hypertension, patients with a history of smoking, and patients with a family history of disease, and the patient's cardiovascular and cerebrovascular data includes the patient's name, age, medical history, and CTA image of the patient;

[0012] Step S102: dividing the patient's cardiovascular and cerebrovascular data into patient's head and neck vascular data and patient's arterial vascular data, and storing them in a head and neck vascular image data sub-library and an arterial vascular image data sub-library respectively.

[0013] As a preferred solution of the cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition described in the present invention, the step S2 specifically includes the following steps:

[0014] Step S201, performing image preprocessing on the patient's cardiovascular data to obtain three-dimensional cardiovascular data, and dividing the three-dimensional cardiovascular data into an image training set and an image test set;

[0015] Step S202, classifying and feature-labeling the three-dimensional vascular images in the image training set, and inputting the class-labeled and feature-labeled image training set into a first convolutional neural network for training, wherein the three-dimensional vascular images are used as data input, and the class-labeled plaque categories, feature-labeled plaque features, and vascular stenosis degrees are used as data output, to obtain a vulnerable plaque recognition model;

[0016] As a preferred solution of the cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition described in the present invention, the step S3 specifically includes the following steps:

[0017] Step S301, gridding the three-dimensional blood vessel image to obtain a three-dimensional fluid simulation model, obtaining vulnerable plaque information according to the vulnerable plaque type through the three-dimensional fluid simulation model, setting resting blood flow parameters and congestive blood flow parameters according to the vulnerable plaque information, inputting the resting blood flow parameters into the three-dimensional fluid simulation model to obtain resting blood flow information, inputting the congestive blood flow parameters into the three-dimensional fluid simulation model to obtain congestive blood flow information, and obtaining a blood flow reserve coefficient through the resting blood flow information and the congestive blood flow information, wherein the blood flow parameters include a blood vessel inlet flow parameter, an outlet flow resistance parameter, blood physical property parameters, and a vessel wall physical property parameter;

[0018] Step S302, obtaining the plaque blood flow abnormality coefficient through the resting blood flow information, the congestion blood flow information and the blood flow reserve coefficient;

[0019] As a preferred solution of the cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition described in the present invention, the step S4 specifically includes the following steps:

[0020] Step S401, obtaining plaque type, plaque characteristics and degree of vascular stenosis of cardiovascular and cerebrovascular data of each group of patients through a plaque recognition model, and obtaining resting blood flow information and blood flow reserve coefficient of cardiovascular and cerebrovascular data of each group of patients through a three-dimensional fluid simulation model;

[0021] Step S402, the doctor marks the rupture cause and warning information of vulnerable plaques according to the plaque type, plaque characteristics, vascular stenosis degree, resting blood flow information and blood flow reserve coefficient of each group of patients' cardiovascular and cerebrovascular data;

[0022] Step S403, inputting the annotated plaque type, plaque characteristics, degree of vascular stenosis, resting blood flow information and blood flow reserve coefficient into a second convolutional neural network for training, wherein the plaque type, plaque characteristics, degree of vascular stenosis, resting blood flow information and blood flow reserve coefficient are used as inputs, and the rupture cause and warning information are used as outputs, to establish a second convolutional neural network model, wherein the first convolutional neural network model and the second convolutional neural network model are jointly constructed as a warning model;

[0023] Step S404, inputting the image test set into the early warning model, using the first convolutional neural network model to obtain the second plaque category, the second plaque feature, and the second vascular stenosis degree of the image test set, using the second convolutional neural network model to obtain the second rupture cause and the second early warning information of the image test set, and obtaining the diagnostic information of the image test set according to the patient's cardiovascular and cerebrovascular data, and using keyword extraction to obtain the first plaque category, the first plaque feature, the first vascular stenosis degree, the first rupture cause and the first early warning information of the image test set according to the diagnostic information, and obtaining the first fit of the first convolutional neural network model according to the first plaque category, the first plaque feature and the first vascular stenosis degree, and adjusting the model parameters of the first convolutional neural network model according to the first fit until the first fit reaches the first expected threshold, obtaining the second fit of the second convolutional neural network model according to the first rupture cause and the first early warning information, and obtaining and adjusting the second convolutional neural network model according to the second fit until the second fit reaches the second expected threshold;

[0024] As a preferred embodiment of the cardiovascular early warning method based on vulnerable plaque image recognition described in the present invention, the vulnerable plaque information includes plaque location, plaque sclerosis volume ratio, plaque fiber cap thickness, blood vessel entrance diameter and blood vessel exit diameter;

[0025] The resting blood flow information includes blood flow time, resting blood flow volume, resting blood flow velocity and resting blood vessel wall shear force;

[0026] The hyperemic blood flow information includes blood flow time, total amount of hyperemic blood flow, hyperemic blood flow velocity and hyperemic blood vessel wall shear force;

[0027] As a preferred solution of the cardiovascular early warning method based on vulnerable plaque image recognition described in the present invention, the image preprocessing of the patient's cardiovascular data to obtain three-dimensional cardiovascular data includes:

[0028] Extracting the patient's CTA image from the patient's cardiovascular and cerebrovascular data, using image segmentation technology to obtain a vascular segmentation image of the patient's CTA image, using a bilateral filter to remove image noise and non-connected blood vessels from the vascular segmentation image according to threshold operation and morphological operation, and using a scale calculation formula to perform image enhancement to obtain scale vascular images of various scales, and establishing a three-dimensional vascular image based on the scale vascular image;

[0029] As a preferred solution of the cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition described in the present invention, wherein: the plaque categories include calcified plaques, non-calcified plaques and mixed plaques;

[0030] The plaque characteristics include low-attenuation plaque, napkin ring sign, punctate calcification and positive remodeling;

[0031] The degree of vascular stenosis includes mild, moderate, severe and high risk;

[0032] As a preferred solution of the cardiovascular early warning method based on vulnerable plaque image recognition described in the present invention, wherein: obtaining the abnormal plaque blood flow coefficient through resting blood flow information, congestive blood flow information and blood flow reserve coefficient includes: ;

[0033] in is the plaque blood flow abnormality coefficient, is the blood flow reserve coefficient, is a constant, is the number of abnormal values ​​in the static blood flow information and the hyperemic blood flow information, is a constant;

[0034] As a preferred solution of the cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition described in the present invention, the scale calculation formula is as follows:

[0035] ;

[0036] in, is the scale transformation function, is the original scale of the blood vessel segmentation image, are different scales of the blood vessel segmentation image, and exp() is an exponential operation with a constant e as the base.

[0037] Beneficial effects of the present invention: The present invention establishes a vascular imaging database for high-risk groups of cardiovascular and cerebrovascular diseases, and establishes a vulnerable plaque recognition model based on the vascular imaging database, which is used to obtain the plaque type, plaque characteristics and plaque stenosis degree of vulnerable plaques through CTA images of patients in the high-risk group, which is conducive to directly and conveniently obtaining detailed data information on vulnerable plaques, saving manpower and material resources for diagnosis and treatment.

[0038] A three-dimensional fluid simulation model is established based on the three-dimensional vascular images of high-risk groups, and the three-dimensional fluid simulation model is used to obtain the resting blood flow information, congestion blood flow information and blood flow reserve coefficient of the blood vessels where the vulnerable plaques are located. This will help medical staff further understand the impact of vulnerable plaques on blood circulation, assess the risk of vascular rupture, and improve the accuracy of clinical diagnosis and the precision of treatment.

[0039] The second convolutional neural network is trained based on the vulnerable plaque recognition model and blood flow information. The rupture cause and warning information are obtained through the plaque type, plaque characteristics, plaque stenosis degree and blood flow information of the vulnerable plaque, providing doctors with accurate and powerful reference for the diagnosis and treatment of vulnerable plaques, which is conducive to the early detection of high-risk groups for cardiovascular and cerebrovascular diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic diagram of the basic flow of a cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0041] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0042] Example 1, reference Figure 1 , is an embodiment of the present invention, and provides a cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition, comprising the following steps:

[0043] Step S1, using big data technology to obtain the patient's cardiovascular and cerebrovascular data from the medical sharing platform, and establishing a vascular imaging database based on the patient's cardiovascular and cerebrovascular data;

[0044] Step S2, performing image preprocessing and data annotation on the patient's cardiovascular data to obtain three-dimensional cardiovascular data, and establishing a vulnerable plaque recognition model using a first convolutional neural network based on the three-dimensional cardiovascular data;

[0045] Step S3, extracting a three-dimensional vascular image from the three-dimensional cardiovascular and cerebrovascular data, establishing a fluid simulation model for the three-dimensional vascular image, and obtaining blood flow information using the fluid simulation model, wherein the blood flow information includes resting blood flow information, hyperemic blood flow information, and a blood flow reserve coefficient;

[0046] Step S4, obtaining rupture causes and warning information according to the output of the vulnerable plaque recognition model and the blood flow information, and inputting the rupture causes and the warning information into a second convolutional neural network model to construct a warning model.

[0047] The vulnerable plaques are those that are unstable and have a tendency to thrombosis;

[0048] The three-dimensional fluid simulation model is used to "virtually" conduct experiments on a computer to simulate actual fluid flow conditions;

[0049] Convolutional neural network is a type of feedforward neural network with deep structure and convolutional calculation. It is one of the representative algorithms of deep learning.

[0050] In this embodiment, a vascular imaging database is established for the high-risk population for cardiovascular and cerebrovascular diseases, and a vulnerable plaque recognition model is established based on the vascular imaging database, which is used to obtain the plaque type, plaque characteristics and degree of plaque stenosis of vulnerable plaques through CTA images of patients in the high-risk population, which is conducive to directly and conveniently obtaining detailed data information on vulnerable plaques, saving manpower and material resources for diagnosis and treatment.

[0051] A three-dimensional fluid simulation model is established based on the three-dimensional vascular images of high-risk groups, and the three-dimensional fluid simulation model is used to obtain the resting blood flow information, congestion blood flow information and blood flow reserve coefficient of the blood vessels where the vulnerable plaques are located. This will help medical staff further understand the impact of vulnerable plaques on blood circulation, assess the risk of vascular rupture, and improve the accuracy of clinical diagnosis and the precision of treatment.

[0052] The second convolutional neural network is trained based on the vulnerable plaque recognition model and blood flow information. The rupture cause and warning information are obtained through the plaque type, plaque characteristics, plaque stenosis degree and blood flow information of the vulnerable plaque, providing doctors with accurate and powerful reference for the diagnosis and treatment of vulnerable plaques, which is conducive to the early detection of high-risk groups for cardiovascular and cerebrovascular diseases.

[0053] The step S1 specifically includes the following steps:

[0054] Step S101, using big data technology to obtain cardiovascular and cerebrovascular data of patients in a high-risk group from a medical sharing platform according to disease tags, wherein the high-risk group includes patients with high-risk factors for coronary heart disease, patients with high-risk factors for cardiovascular and cerebrovascular atherosclerosis, and patients with acute myocardial infarction, wherein the patients with high-risk factors for coronary heart disease include patients with diabetes, patients with hyperlipidemia, patients with hypertension, patients with a history of smoking, and patients with a family history of disease, and the patient's cardiovascular and cerebrovascular data includes the patient's name, age, medical history, and CTA image of the patient;

[0055] Step S102: dividing the patient's cardiovascular and cerebrovascular data into patient's head and neck vascular data and patient's arterial vascular data, and storing them in a head and neck vascular image data sub-library and an arterial vascular image data sub-library respectively.

[0056] The disease label is a label used to identify the disease type of the patient;

[0057] The CTA image, also called CT angiography, combines CT enhancement technology with thin-layer, large-range, and rapid scanning technology to clearly display the details of blood vessels in various parts of the body through reasonable post-processing.

[0058] In this embodiment, the high-risk population obtained includes 4139 people, the patients with high-risk factors for coronary heart disease include 3114 people, the patients with high-risk factors for cardiovascular and cerebrovascular atherosclerosis include 2008 people, and the patients with acute myocardial infarction include 443 people.

[0059] In this embodiment, big data technology is used to obtain cardiovascular data of patients in high-risk groups for cardiovascular and cerebrovascular diseases from a medical sharing platform, providing comprehensive, accurate and specific data support for establishing vulnerable plaque identification models and three-dimensional fluid simulation models, thereby ensuring the authenticity of the original data.

[0060] The step S2 specifically includes the following steps:

[0061] Step S201, performing image preprocessing on the patient's cardiovascular data to obtain three-dimensional cardiovascular data, and dividing the three-dimensional cardiovascular data into an image training set and an image test set;

[0062] Step S202, classifying and feature labeling the three-dimensional vascular images in the image training set, and inputting the class-labeled and feature-labeled image training set into the first convolutional neural network for training, wherein the three-dimensional vascular images are used as data input, and the class-labeled plaque categories, feature-labeled plaque features, and the degree of vascular stenosis are used as data output, so as to obtain a vulnerable plaque recognition model.

[0063] In this embodiment, three-dimensional cardiovascular and cerebrovascular data are obtained based on the patient's cardiovascular and cerebrovascular data, and a vulnerable plaque recognition model is established using three-dimensional vascular images to obtain the plaque type, plaque characteristics and degree of vascular stenosis of the vulnerable plaque. This provides simplicity and convenience for a detailed understanding of the vulnerable plaque, saves manpower and material resources, and provides an accurate diagnostic basis for subsequent acquisition of rupture causes and warning information based on the detailed situation of the vulnerable plaque.

[0064] The step S3 specifically comprises the following steps:

[0065] Step S301, gridding the three-dimensional blood vessel image to obtain a three-dimensional fluid simulation model, obtaining vulnerable plaque information according to the vulnerable plaque type through the three-dimensional fluid simulation model, setting resting blood flow parameters and congestive blood flow parameters according to the vulnerable plaque information, inputting the resting blood flow parameters into the three-dimensional fluid simulation model to obtain resting blood flow information, inputting the congestive blood flow parameters into the three-dimensional fluid simulation model to obtain congestive blood flow information, and obtaining a blood flow reserve coefficient through the resting blood flow information and the congestive blood flow information, wherein the blood flow parameters include a blood vessel inlet flow parameter, an outlet flow resistance parameter, blood physical property parameters, and a vessel wall physical property parameter;

[0066] Step S302, obtaining the plaque blood flow abnormality coefficient through the resting blood flow information, the congestive blood flow information and the blood flow reserve coefficient.

[0067] In this embodiment, by establishing a three-dimensional fluid simulation model, obtaining the resting blood flow information, congestion blood flow information and blood flow reserve coefficient of the blood vessel where the vulnerable plaque is located, it is beneficial to further understand the impact of the vulnerable plaque on the blood circulation of the blood vessel in detail, and provide specific and comprehensive data support for the subsequent acquisition of the causes and early warning information of blood vessel rupture.

[0068] The step S4 specifically comprises the following steps:

[0069] Step S401, obtaining plaque type, plaque characteristics and degree of vascular stenosis of cardiovascular and cerebrovascular data of each group of patients through a plaque recognition model, and obtaining resting blood flow information and blood flow reserve coefficient of cardiovascular and cerebrovascular data of each group of patients through a three-dimensional fluid simulation model;

[0070] Step S402, the doctor marks the rupture cause and warning information of vulnerable plaques according to the plaque type, plaque characteristics, vascular stenosis degree, resting blood flow information and blood flow reserve coefficient of each group of patients' cardiovascular and cerebrovascular data;

[0071] Step S403, inputting the annotated plaque type, plaque characteristics, degree of vascular stenosis, resting blood flow information and blood flow reserve coefficient into a second convolutional neural network for training, wherein the plaque type, plaque characteristics, degree of vascular stenosis, resting blood flow information and blood flow reserve coefficient are used as inputs, and the rupture cause and warning information are used as outputs, to establish a second convolutional neural network model, wherein the first convolutional neural network model and the second convolutional neural network model are jointly constructed as a warning model;

[0072] Step S404, input the image test set into the early warning model, use the first convolutional neural network model to obtain the second plaque category, the second plaque feature and the second vascular stenosis degree of the image test set, use the second convolutional neural network model to obtain the second rupture cause and the second early warning information of the image test set, and obtain the diagnostic information of the image test set according to the patient's cardiovascular and cerebrovascular data, and use keyword extraction to obtain the first plaque category, the first plaque feature, the first vascular stenosis degree, the first rupture cause and the first early warning information of the image test set according to the diagnostic information, and obtain the first fit of the first convolutional neural network model according to the first plaque category, the first plaque feature and the first vascular stenosis degree, and adjust the model parameters of the first convolutional neural network model according to the first fit until the first fit reaches the first expected threshold, obtain the second fit of the second convolutional neural network model according to the first rupture cause and the first early warning information, and obtain and adjust the second convolutional neural network model according to the second fit until the second fit reaches the second expected threshold.

[0073] In this embodiment, the first expected threshold is 95%, and the second expected threshold is 93%.

[0074] In this embodiment, the output of the vulnerable plaque recognition model and blood flow information are input into the second convolutional neural network model, and the rupture cause and warning information are used as output to construct a warning model. The image test set is used to adjust the parameters of the first convolutional neural network model and the second convolutional neural network model, which provides a powerful reference for risk warning and diagnosis and treatment of vulnerable plaques based on the patient's CTA images, and ensures the accuracy of risk warning.

[0075] The vulnerable plaque information includes plaque location, plaque sclerosis volume ratio, plaque fiber cap thickness, blood vessel entrance diameter and blood vessel exit diameter;

[0076] The resting blood flow information includes blood flow time, resting blood flow volume, resting blood flow velocity and resting blood vessel wall shear force;

[0077] The hyperemic blood flow information includes blood flow time, total amount of hyperemic blood flow, hyperemic blood flow velocity and hyperemic blood vessel wall shear force.

[0078] In this embodiment, the use of a three-dimensional fluid simulation model to obtain blood flow information is helpful for medical staff to further understand the impact of vulnerable plaques on blood circulation and assess the risk of blood vessel rupture, and provides a strong reference for risk warning and diagnosis and treatment of vulnerable plaques based on the patient's CTA images.

[0079] Image preprocessing of the patient's cardiovascular data to obtain three-dimensional cardiovascular data includes:

[0080] The patient's CTA image is extracted from the patient's cardiovascular and cerebrovascular data, and a vascular segmentation image of the patient's CTA image is obtained using image segmentation technology. A bilateral filter is used to remove image noise and non-connected blood vessels from the vascular segmentation image according to threshold operation and morphological operation. A scale calculation formula is used to perform image enhancement to obtain scale vascular images of various scales, and a three-dimensional vascular image is established based on the scale vascular image.

[0081] In this embodiment, the bilateral filter is used to protect the edge characteristics of the image during the image denoising process;

[0082] The threshold operation refers to the process of setting a certain threshold and uniformly processing pixels greater than the threshold or pixels less than the threshold;

[0083] The morphological operation is a shape-based image processing technology that changes the shape and characteristics of an image by performing specific operations on structural elements and images;

[0084] Image enhancement is a process of purposefully emphasizing the overall or local characteristics of an image, making the image clearer, enlarging the differences between features of different objects in the image, suppressing uninteresting features, improving image quality, and enriching the amount of information.

[0085] In this embodiment, image preprocessing is performed on the patient CTA images in the cardiovascular data to obtain three-dimensional vascular images, which provide accurate and detailed image basis for establishing a three-dimensional fluid simulation model, thereby ensuring the authenticity and accuracy of the three-dimensional fluid simulation model.

[0086] The plaque categories include calcified plaque, non-calcified plaque and mixed plaque;

[0087] The plaque characteristics include low-attenuation plaque, napkin ring sign, punctate calcification and positive remodeling;

[0088] The degree of vascular stenosis includes mild, moderate, severe and high risk.

[0089] In this embodiment, the vulnerable plaque recognition model is used to obtain the plaque type, plaque characteristics and degree of vascular stenosis of the vulnerable plaque, which is conducive to directly and conveniently obtaining detailed data information of the vulnerable plaque, saving manpower and material resources for diagnosis and treatment, and providing a strong reference for risk warning and diagnosis and treatment of vulnerable plaques based on CTA images, thereby ensuring the accuracy of risk warning.

[0090] Obtaining the abnormal plaque blood flow coefficient through resting blood flow information, hyperemia blood flow information and blood flow reserve coefficient includes: ;

[0091] in is the plaque blood flow abnormality coefficient, is the blood flow reserve coefficient, is a constant, is the number of abnormal values ​​in the static blood flow information and the hyperemic blood flow information, is a constant.

[0092] In this embodiment, obtaining the plaque blood flow abnormality coefficient through resting blood flow information, congestive blood flow information and blood flow reserve coefficient is conducive to further understanding the impact of vulnerable plaques on vascular blood circulation, and provides specific and comprehensive data support for subsequent acquisition of vascular rupture inducements and warning information.

[0093] The scale calculation formula is as follows:

[0094] ;

[0095] in, is the scale transformation function, is the original scale of the blood vessel segmentation image, are different scales of the blood vessel segmentation image, and exp() is an exponential operation with a constant e as the base.

[0096] In this embodiment, a scale calculation formula is used to perform image enhancement to obtain scale vascular images of various scales, and a three-dimensional vascular image is established based on the scale vascular image, providing an accurate and detailed image basis for establishing a three-dimensional fluid simulation model, thereby ensuring the authenticity and accuracy of the three-dimensional fluid simulation model.

[0097] The present invention establishes a vascular imaging database for high-risk groups of cardiovascular and cerebrovascular diseases, and establishes a vulnerable plaque recognition model based on the vascular imaging database, which is used to obtain the plaque type, plaque characteristics and plaque stenosis degree of vulnerable plaques through CTA images of patients in the high-risk group, which is conducive to directly and conveniently obtaining detailed data information on vulnerable plaques, saving manpower and material resources for diagnosis and treatment.

[0098] A three-dimensional fluid simulation model is established based on the three-dimensional vascular images of high-risk groups, and the three-dimensional fluid simulation model is used to obtain the resting blood flow information, congestion blood flow information and blood flow reserve coefficient of the blood vessels where the vulnerable plaques are located. This will help medical staff further understand the impact of vulnerable plaques on blood circulation, assess the risk of vascular rupture, and improve the accuracy of clinical diagnosis and the precision of treatment.

[0099] The second convolutional neural network is trained based on the vulnerable plaque recognition model and blood flow information. The rupture cause and warning information are obtained through the plaque type, plaque characteristics, plaque stenosis degree and blood flow information of the vulnerable plaque, providing doctors with accurate and powerful reference for the diagnosis and treatment of vulnerable plaques, which is conducive to the early detection of high-risk groups for cardiovascular and cerebrovascular diseases.

[0100] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program codes. Among them, the storage medium may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0101] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition, characterized in that: The following steps are involved: Step S1, using big data technology to obtain the patient's cardiovascular and cerebrovascular data from the medical sharing platform, and establishing a vascular imaging database based on the patient's cardiovascular and cerebrovascular data; Step S2, performing image preprocessing and data annotation on the patient's cardiovascular data to obtain three-dimensional cardiovascular data, and establishing a vulnerable plaque recognition model using a first convolutional neural network based on the three-dimensional cardiovascular data; Step S3, extracting a three-dimensional vascular image from the three-dimensional cardiovascular and cerebrovascular data, establishing a fluid simulation model for the three-dimensional vascular image, and obtaining blood flow information using the fluid simulation model, wherein the blood flow information includes resting blood flow information, hyperemic blood flow information, and a blood flow reserve coefficient; Step S4, obtaining rupture causes and warning information according to the output of the vulnerable plaque recognition model and the blood flow information, and inputting the rupture causes and the warning information into a second convolutional neural network model to construct a warning model; The step S4 specifically comprises the following steps: Step S401, obtaining plaque type, plaque characteristics and degree of vascular stenosis of cardiovascular and cerebrovascular data of each group of patients through a plaque recognition model, and obtaining resting blood flow information and blood flow reserve coefficient of cardiovascular and cerebrovascular data of each group of patients through a three-dimensional fluid simulation model; Step S402, the doctor marks the rupture cause and warning information of vulnerable plaques according to the plaque type, plaque characteristics, vascular stenosis degree, resting blood flow information and blood flow reserve coefficient of each group of patients' cardiovascular and cerebrovascular data; Step S403, inputting the annotated plaque type, plaque characteristics, degree of vascular stenosis, resting blood flow information and blood flow reserve coefficient into a second convolutional neural network for training, wherein the plaque type, plaque characteristics, degree of vascular stenosis, resting blood flow information and blood flow reserve coefficient are used as inputs, and the rupture cause and warning information are used as outputs, to establish a second convolutional neural network model, wherein the first convolutional neural network model and the second convolutional neural network model are jointly constructed as a warning model; Step S404, input the image test set into the early warning model, use the first convolutional neural network model to obtain the second plaque category, the second plaque feature and the second vascular stenosis degree of the image test set, use the second convolutional neural network model to obtain the second rupture cause and the second early warning information of the image test set, and obtain the diagnostic information of the image test set according to the patient's cardiovascular and cerebrovascular data, and use keyword extraction to obtain the first plaque category, the first plaque feature, the first vascular stenosis degree, the first rupture cause and the first early warning information of the image test set according to the diagnostic information, and obtain the first fit of the first convolutional neural network model according to the first plaque category, the first plaque feature and the first vascular stenosis degree, and adjust the model parameters of the first convolutional neural network model according to the first fit until the first fit reaches the first expected threshold, obtain the second fit of the second convolutional neural network model according to the first rupture cause and the first early warning information, and obtain and adjust the second convolutional neural network model according to the second fit until the second fit reaches the second expected threshold.

2. The cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition according to claim 1, characterized in that: The step S1 specifically includes the following steps: Step S101, using big data technology to obtain cardiovascular and cerebrovascular data of patients in a high-risk group from a medical sharing platform according to disease tags, wherein the high-risk group includes patients with high-risk factors for coronary heart disease, patients with high-risk factors for cardiovascular and cerebrovascular atherosclerosis, and patients with acute myocardial infarction, wherein the patients with high-risk factors for coronary heart disease include patients with diabetes, patients with hyperlipidemia, patients with hypertension, patients with a history of smoking, and patients with a family history of disease, and the patient's cardiovascular and cerebrovascular data includes the patient's name, age, medical history, and CTA image of the patient; Step S102: dividing the patient's cardiovascular and cerebrovascular data into patient's head and neck vascular data and patient's arterial vascular data, and storing them in a head and neck vascular image data sub-library and an arterial vascular image data sub-library respectively.

3. The cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition according to claim 1, characterized in that: The step S2 specifically includes the following steps: Step S201, performing image preprocessing on the patient's cardiovascular data to obtain three-dimensional cardiovascular data, and dividing the three-dimensional cardiovascular data into an image training set and an image test set; Step S202, classifying and feature labeling the three-dimensional vascular images in the image training set, and inputting the class-labeled and feature-labeled image training set into the first convolutional neural network for training, wherein the three-dimensional vascular images are used as data input, and the class-labeled plaque categories, feature-labeled plaque features, and the degree of vascular stenosis are used as data output, so as to obtain a vulnerable plaque recognition model.

4. The cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition according to claim 1, characterized in that: The step S3 specifically comprises the following steps: Step S301, gridding the three-dimensional blood vessel image to obtain a three-dimensional fluid simulation model, obtaining vulnerable plaque information according to the vulnerable plaque type through the three-dimensional fluid simulation model, setting resting blood flow parameters and congestive blood flow parameters according to the vulnerable plaque information, inputting the resting blood flow parameters into the three-dimensional fluid simulation model to obtain resting blood flow information, inputting the congestive blood flow parameters into the three-dimensional fluid simulation model to obtain congestive blood flow information, and obtaining a blood flow reserve coefficient through the resting blood flow information and the congestive blood flow information, wherein the blood flow parameters include a blood vessel inlet flow parameter, an outlet flow resistance parameter, blood physical property parameters, and a vessel wall physical property parameter; Step S302, obtaining the plaque blood flow abnormality coefficient through the resting blood flow information, the congestive blood flow information and the blood flow reserve coefficient.

5. The cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition according to claim 4, characterized in that: The vulnerable plaque information includes plaque location, plaque sclerosis volume ratio, plaque fiber cap thickness, blood vessel entrance diameter and blood vessel exit diameter; The resting blood flow information includes blood flow time, resting blood flow volume, resting blood flow velocity and resting vascular wall shear force; The hyperemic blood flow information includes blood flow time, total amount of hyperemic blood flow, hyperemic blood flow velocity and hyperemic blood vessel wall shear force.

6. The cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition according to claim 3, characterized in that: Image preprocessing of the patient's cardiovascular data to obtain three-dimensional cardiovascular data includes: The patient's CTA image is extracted from the patient's cardiovascular and cerebrovascular data, and a vascular segmentation image of the patient's CTA image is obtained using image segmentation technology. A bilateral filter is used to remove image noise and non-connected blood vessels from the vascular segmentation image according to threshold operation and morphological operation. A scale calculation formula is used to perform image enhancement to obtain scale vascular images of various scales, and a three-dimensional vascular image is established based on the scale vascular image.

7. The cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition according to claim 3, characterized in that: The plaque categories include calcified plaque, non-calcified plaque and mixed plaque; The plaque characteristics include low-attenuation plaque, napkin ring sign, punctate calcification and positive remodeling; The degree of vascular stenosis includes mild, moderate, severe and high risk.

8. The cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition according to claim 4, characterized in that: Obtaining the abnormal plaque blood flow coefficient through resting blood flow information, hyperemia blood flow information and blood flow reserve coefficient includes: ; in is the plaque blood flow abnormality coefficient, is the blood flow reserve coefficient, is a constant, is the number of abnormal values ​​in the static blood flow information and the hyperemic blood flow information, is a constant.

9. The cardiovascular and cerebrovascular early warning method based on vulnerable plaque image recognition according to claim 6, characterized in that: The scale calculation formula is as follows: ； in, is the scale transformation function, is the original scale of the blood vessel segmentation image, are different scales of the blood vessel segmentation image, and exp() is an exponential operation with a constant e as the base.

Images