CN117671145A

CN117671145A - Human body blood vessel digital twin method, system and equipment based on real world research

Info

Publication number: CN117671145A
Application number: CN202311650790.3A
Authority: CN
Inventors: 鞠悦; 杜伯仁
Original assignee: Shanghai Wanyi Medical Technology Co ltd
Current assignee: Shanghai Wanyi Medical Technology Co ltd
Priority date: 2023-12-05
Filing date: 2023-12-05
Publication date: 2024-03-08

Abstract

Human body blood vessel digital twin method, system and equipment based on real world research comprise: acquiring medical images of a target blood vessel region, dividing image frames according to an image frame time sequence, and automatically extracting the blood vessels from the frame images through a blood vessel dividing model; performing vessel 3D modeling according to the vessel attribute data; dividing the medical image into a plurality of blocks, dividing each block into a plurality of blood vessel video sequences, dividing each blood vessel video sequence into video blocks at equal intervals, obtaining the blood flow velocity of each blood vessel of the video blocks according to time sequence association among frames in the video blocks, mapping the blood flow velocity into corresponding blood vessels in a blood vessel 3D model to construct blood flow information of each blood vessel in the blood vessel 3D model corresponding to the medical image, fitting through a pre-hydrodynamic model to realize more real blood flow information, and then carrying out blood vessel plaque parameter model identification on the frame images of a target blood vessel region to map into the corresponding blood vessels of the blood vessel 3D model. The twinning of the human blood vessels can help doctors to more accurately carry out clinical diagnosis and the selection of treatment schemes.

Description

Human body blood vessel digital twin method, system and equipment based on real world research

Technical Field

The invention relates to the field of artificial intelligence, in particular to a human body blood vessel digital twin method, system and equipment based on real world research.

Background

Digital Twin (DT) is a simulation process that fully utilizes data such as actual models, sensor updates, running histories, and the like, and cooperates with multiple disciplines, multiple actual amounts, multiple scales, and multiple probabilities, and completes association in a privacy space, thereby reflecting the full life cycle process of corresponding object equipment. DT may be considered a digital association system of one or more important, mutually dependent equipment systems. DT is a universally adapted theoretical technology system and can be applied in a plurality of fields, and in recent years, as DT technology is applied in medicine, it will become possible to provide personalized diagnosis and treatment for patients, which are personalized accurate medical treatment. Some previous researches take the transformation of computing physiology into clinical practice as a hope, and a digital technology is utilized to generate a privacy physiological human body, and a certain progress is made.

The development of technologies such as big data, cloud computing, privacy reality, the Internet of things and the like enables us to obtain more data more easily, lays a foundation for the application of DT technology, and provides finer dimensions for clinicians and researchers to study the occurrence and development of diseases and to perform more accurate diagnosis and treatment. The model created by DT is a privacy copy of human organs, tissues, cells or microenvironment, is continuously adjusted according to the change of online data, and can predict the future of corresponding objects and indexes. Meanwhile, the DT is not only a digital model, but also a living, intelligent and continuously developed model, and the future state can be continuously predicted while optimizing the flow and algorithm according to closed-loop optimization between the DT and the surrounding environment.

The Sichuan medical college discloses 202311048018.4 a prediction method and a system based on association of digital twin sepsis patients, first detection data of the sepsis patients to be detected are obtained, and illness state clustering is carried out on the sepsis patients to be detected, so that illness state clustering results are obtained; acquiring second detection data of patients to be detected with sepsis under the condition aggregation sets of all patients by adopting different data detection strategies; generating a sepsis digital twin simulation model, and inputting second detection data to obtain pre-alarm data of a patient to be detected sepsis; acquiring treatment archive data of the discharged rehabilitation patient, extracting detection data of the rehabilitation patient and pre-alarm data for comparison, and obtaining an alarm comparison result; and sending risk alarm predictions to different sepsis patient terminals by adopting different alarm strategies. The invention can effectively extract and analyze data according to different strategies of the damaged part of the sepsis patient, and combines the historical rehabilitation patient data to perform early warning comparison, thereby improving the correlation predictability of the sepsis.

In clinical diagnosis and treatment processes, human body blood vessels of patients need to be simulated, so that doctors can be helped to more accurately conduct clinical diagnosis and treatment scheme selection. For example, vascular lesions can cause a number of diseases, such as stroke and coronary heart disease. Because of the difficulty in directly researching blood vessels in a human body, the deep cause of vascular lesions is hardly known, so that the creation of a proper and similar blood vessel research model outside the human body is important for the development of the whole field. However, there is currently no human vascular digital twinning system and apparatus based on real world research.

Disclosure of Invention

The invention provides a human body blood vessel digital twin method, a system and equipment based on real world research, which are used for solving the technical problems that the prior art cannot truly simulate the human body blood vessel of a patient, and is not beneficial to doctors to more accurately perform clinical diagnosis and selection of treatment schemes.

Drawings

FIG. 1 is a diagram of a human vascular digital twinning methodology based on real world research;

fig. 2 is a schematic diagram of a human vascular digital twin system based on real world research.

Detailed Description

The present invention will be described in detail below with reference to the accompanying drawings.

In embodiments of the present disclosure, the term "model" is capable of processing an input and providing a corresponding output. Taking the neural network model as an example, it generally includes an input layer, an output layer, and one or more hidden layers between the input layer and the output layer. Models used in deep learning applications (also referred to as "deep learning models") typically include many hidden layers, thereby extending the depth of the network. The layers of the neural network model are connected in sequence such that the output of the previous layer is used as the input of the subsequent layer, wherein the input layer receives the input of the neural network model and the output of the output layer is the final output of the neural network model. Each layer of the neural network model includes one or more nodes (also referred to as processing nodes or neurons), each of which processes inputs from a previous layer. The terms "neural network," "model," "network," and "neural network model" are used interchangeably herein.

Referring to fig. 1, a flowchart of a human vascular digital twin method based on real world research is shown. It comprises the following steps:

s110, acquiring medical images of a target blood vessel region, dividing image frames according to an image frame time sequence, and automatically extracting the blood vessels from the frame images through a blood vessel dividing model;

s120, carrying out blood vessel 3D modeling according to the attribute data comprising the blood vessels;

s130, dividing the medical image into a plurality of blocks, dividing each block into a plurality of blood vessel video sequences, dividing each blood vessel video sequence into video blocks at equal intervals, obtaining the blood flow velocity of each blood vessel of the video blocks according to the time sequence association among frames in the video blocks, and mapping the blood flow velocity into corresponding blood vessels in a blood vessel 3D model so as to construct blood flow information of each blood vessel in the blood vessel 3D model corresponding to the medical image;

and S140, carrying out vascular plaque parameter model identification on the frame images of the target vascular region so as to map the images into corresponding blood vessels of the vascular 3D model.

And establishing a three-dimensional digital twin system of the human blood vessel based on the feature structure of the real human, wherein the three-dimensional digital twin system comprises a three-dimensional patient model constructed by a patient, and the three-dimensional patient model comprises a plurality of key point information associated with the human body part. The three-dimensional model of the patient can embody the condition that the actual blood vessel position, the body shape and the like of the patient are consistent, and the three-dimensional digital twin system established based on the live-action image can reflect the dynamic change of the blood vessel of the patient personnel in real time and accurately. Especially has good effects in preoperative planning, early warning of symptoms and the like.

Considering the complexity of human blood vessels, the application can only construct a certain part of the human body, such as brain, lower limbs and lung blood vessel digital twin. The digital twin system for erecting the whole human blood vessel can be initially constructed. In view of the characteristics of blood vessels, it is also common to construct digital twin systems or devices for venous blood vessels and arteries of the human body.

The application firstly takes human body vascular digital twin at a certain part as an example, and specifically describes each step.

1. Step S110:

and acquiring a medical image of the target blood vessel region. In this step, the target vessel region may be a coronary vessel region, and the medical image may be acquired by the image acquisition apparatus. The medical image may be DICOM (Digital Imaging and Communications in Medicine), i.e. digital imaging and communication in medicine, is an international standard for medical images and related information, defining a medical image format with quality that satisfies clinical requirements and that can be used for data exchange. Medical images for which plaque analysis may follow-up include, but are not limited to: CTA, IVUS, OCT; medical images for which a vascular model may be subsequently constructed include, but are not limited to: CTA, MRI, DSA. The contrast image to be segmented may be a Magnetic Resonance Angiography (MRA) image or a TOF (Time of Flight) magnetic resonance angiography image. In practical application, the mode of the contrast image to be segmented needs to be determined according to the practical application situation, and the application is not limited in this respect. And constructing an initial blood vessel model of the target blood vessel region based on the medical image. In practice, the initial vessel model may be constructed based on medical images of the target vessel region using prior art techniques. Taking CTA as an example, the gray value of a blood vessel region is obviously different from that of a non-blood vessel region, and the blood vessel can be judged according to the gray value of a photographed medical image, so that the blood vessel is segmented, and then a blood vessel model is established. The vessel model herein mainly refers to a geometric model, including a single vessel or a plurality of vessels.

In this example the automatic vessel extraction further comprises:

obtaining a plurality of MRA images from the medical image of the target vessel region;

in the preprocessing stage of vessel segmentation, gaussian filtering processing is carried out on each MRA image, then a second derivative of each pixel point is calculated to construct a Hessian matrix, characteristic values including brightness and color are obtained and brought into vessel similarity functions to obtain filter responses under different scales, and when the scales are matched with local structures, a filter response value is obtained, so that whether the current pixel point belongs to a vessel structure is judged;

and extracting pixel points belonging to the vascular structure by using the Hessian RecursiveGaussian ImageFilter, and constructing corresponding vascular information.

MRA (magnetic resonance angiography) images are typically multi-dimensional, such as slice images of a two-dimensional plane or three-dimensional volumetric data, requiring a filtering process for each dimension when applying a gaussian filtering algorithm. Determining the size of the filter (i.e., the value of (k)) and the weight of the gaussian filter requires determining the size of the filter (i.e., the value of (k)) before applying the gaussian filtering algorithm. The choice of filter size affects the filtering effect and usually needs to be adjusted according to the specific situation.

2. Step S120.

The vessel 3D modeling further comprises:

acquiring a three-dimensional central line of a blood vessel contour of a target blood vessel, importing an initial blood vessel model based on the three-dimensional central line, and performing smoothing treatment on the initial blood vessel model to obtain a preliminary three-dimensional blood vessel model of the target blood vessel;

the method comprises the steps of carrying out real simulation on a target blood vessel by using a Snake model, firstly, giving an initial curve to the corresponding blood vessel of the three-dimensional blood vessel model, and enabling the curve to deform in the three-dimensional blood vessel model by using a minimized energy floodfunction and continuously approach to the target contour of the target blood vessel

To obtain a 3D model of the blood vessel.

3. Step 130:

each block is divided into a plurality of blood vessel video sequences, each blood vessel video sequence is divided into video blocks at equal intervals, a corresponding neural network model is trained, and according to time sequence association among frames in the video blocks, obtaining blood flow velocity of each blood vessel of the video block further comprises:

establishing a three-dimensional center line of the medical image, and establishing coordinate mapping comprising the position, the size and the identification of the target blood vessel with a blood vessel 3D model;

dividing the medical image into a plurality of blocks, and mapping each block with block coordinates of a blood vessel 3D model, including the position, the size and the identification of a target blood vessel;

dividing each block into a plurality of blood vessel video sequences, equally dividing each blood vessel video sequence into video blocks at equal intervals, firstly finding out a first frame image, calculating the information entropy of a plurality of neighborhoods around each blood vessel pixel in the image frame, and finding out the pixel position with the maximum information entropy in each blood vessel as a stable point;

taking an ith frame as a reference frame and an (i+1) th frame as a target frame, taking a characteristic region in the reference frame as a tracking target, tracking the position of each blood vessel characteristic region in the target frame by using a correlation filtering algorithm, and recording the offset of each blood vessel characteristic region in the target frame relative to the characteristic region in the reference frame; and (3) traversing all image frames in the image sequence until all offset information of blood vessels corresponding to the blood vessel video sequences is obtained, and obtaining the blood flow velocity information of each blood vessel through the offset.

In general, blood flow velocity varies depending on the blood vessel region, and is specifically as follows:

1. the normal carotid artery flow rate is less than 1.2m/s, the normal carotid artery two-dimensional image is a cross section, the lumen is circular, the longitudinal section scanning tube wall is formed by two parallel light bands, the tube wall consists of an inner membrane, a middle membrane and an outer membrane, the inner membrane has low echo, is slender and smooth, has good continuity, the middle layer is a dark zone, the outer membrane is a blood vessel wall, the outermost layer is a bright light band, and the tube wall thickness is about 1mm;

2. the normal flow rate of the abdominal aorta is less than 1.8m/s, the abdominal aorta is positioned at the left front of the spine, the abdominal aorta is divided into left and right common iliac arteries downwards to the level of the fourth lumbar vertebra, and the inner diameter of the proximal section of the abdominal aorta of a normal adult is 2-3cm;

3. the normal blood flow rate of the inferior vena cava is 5-25cm/s, and the inferior vena cava is located right anterior to the spinal column and ascending along the right side of the abdominal aorta.

When the standard value of a certain blood vessel plus all the offset information is converted into the speed as the expression, the blood flow speed is obtained.

The constructing blood flow information of each blood vessel in the blood vessel 3D model corresponding to the medical image further comprises the following steps:

establishing a reference value of blood flow velocity of different blood vessel parts in advance;

the number of blood vessel flow velocity of the user is collected by ultrasonic, a hydrodynamic blood flow model is initially established,

fitting with blood flow information of a corresponding blood vessel in the blood vessel 3D model, judging that the error between the blood flow information and the reference value is within a preset value, and taking the fitted data as blood flow data of the current blood vessel 3D model.

The method can more truly reflect the realized blood flow velocity of the user and improve the authenticity of the user

Vascular blood flow may refer to the flow of blood in a blood vessel. Hydrodynamic is the discipline of studying the movement and mechanical properties of fluids (including blood), so that the flow of blood can be modeled and analyzed by hydrodynamic methods. The blood flow velocity model is a mathematical model describing the flow behavior of blood in a blood vessel. Common blood flow models include newtonian and non-newtonian blood flow models that can be employed.

Newtonian blood fluid model the blood is assumed to be a Newtonian blood fluid with a constant viscosity. In consideration of the fact that a model which is easy to build is adopted in the early stage of building, for example, different corresponding flow speed ranges of carotid artery flow speed, abdominal aorta and inferior vena cava are adopted, blood flow speed training is carried out according to different positions during training, and in the model, corresponding shear stress and pressure values accepted by blood flow at corresponding positions are detected through a certain means or a sensor. The flow mechanism of blood in the blood vessel can be established according to the Navier-Stokes equation, so that the simulation and the prediction of the blood flow dynamics are optimized.

To solve the Navier-Stokes equation, appropriate boundary conditions and initial conditions need to be given. The boundary condition may include a value of a velocity, pressure, or other physical quantity, and the initial condition is an initial state of the fluid at an initial time. The Navier-Stokes equation can be solved by numerical methods, such as finite difference methods, finite element methods, or Computational Fluid Dynamics (CFD) methods, to obtain the flow velocity and pressure distribution of the blood. The velocity v (x, t) and the pressure value f (x, t) are decomposed into three components in coordinates:

·v(x，t)＝(v ₁ (x，t)，v ₂ (x，t)，v ₃ (x，t)，f(x，t)＝(f ₁ (x，t)，f ₂ (x，t)，f ₃ (x, t)) then the Navier-Stokes equation can be written in the form i=1, 2,3:

·

wherein the pressure p (x, t) is given a constant value, and the position flow velocity v (x, t) of the normal carotid artery takes three values smaller than 1.2m/s, the normal blood flow velocity v (x, t) of the same abdominal aorta takes three values smaller than 1.8m/s, and the blood flow velocity v (x, t) of the inferior vena cava takes three values between 5-25 cm/s. Parameters in the Navier-Stokes equation are fitted by an optimization algorithm (e.g., least squares, gradient descent, etc.) to obtain more accurate flow simulation and prediction.

In optimizing the algorithm, a gradient descent method, which is a common optimization algorithm, is used to solve the problem of minimizing the loss function. In the fitting problem, a gradient descent method may be used to fit parameters in the blood fluid model such that the gap between the predicted result and the actual observed value of the blood fluid model is minimized.

Assume that there is a set of samples ({ (x_i, y_i) } _ { i=1 } n), where (x_i) represents the input,

(y_i) represents an output. Our goal is to find a functional model (f (x; \theta)), where (\theta) represents parameters in the model such that the fitting error of the model to the sample set is minimized.

The fitting error is typically measured by a loss function, a common loss function including Mean Square Error (MSE)

And Cross Entropy (Cross Entropy), etc. Taking mean square error as an example, the loss function can be expressed as:

mse=1/n Σi= 1^n (y_i+_i) ≡2 where n represents the number of samples. y_i represents the actual value, _i represents the predicted value, and Σi represents the sum of all data samples.

The basic idea of the gradient descent method is to update the parameters in an iterative manner so that the loss function is minimized. Specifically, the current parameter value is updated along the negative gradient direction of the loss function during each iteration, and the step size of each iteration is controlled. The gradient refers to the partial derivative of the loss function to the parameter, and the optimal parameter value can be obtained by continuously iterating and updating the parameter until the loss function converges or reaches a certain iteration number, so that an optimal model is obtained. The above is a conventional algorithm, and the main core is to control the iteration times.

The Navier-Stokes equation is obtained through training and solving, and then the position of the parameter is judged, and then the parameter enters a blood flow velocity model corresponding to the corresponding position (normal carotid artery, abdominal artery or other veins) to output the corresponding blood flow velocity.

4. Step S140.

And carrying out plaque analysis on the target blood vessel region based on the medical image of the target blood vessel region to obtain blood vessel plaque parameters. In practice, plaque analysis may be performed based on medical images of the target vessel region using methods known in the art to obtain vessel plaque parameters. The plaque here includes calcified plaque, non-calcified plaque and mixed plaque. Illustratively, taking CTA as an example, the nature of plaque may be determined from the gray value of the medical image taken. Vascular plaque parameters include, but are not limited to: plaque type, plaque location, plaque stenosis, blood vessel wall thickness, calcification score, plaque angle, plaque volume, plaque thickness, reconstitution index, fat attenuation index, whether positive reconstitution, napkin ring sign, punctual calcification, low density plaque, etc.

In this example, if it is a certain part of the human body, only one model training needs to be performed: training an initial model, training plaque parameters including plaque stenosis degree, blood vessel wall thickness and plaque thickness, dividing the plaque parameters of a sample level into a plurality of levels according to the plaque stenosis degree, dividing the blood vessel wall thickness into a plurality of levels, and dividing the plaque thickness into a plurality of levels for indexing and training; carrying out plaque parameter model identification on the frame images of the target blood vessel region to obtain the level information of plaque parameters at the corresponding positions of the blood vessel; and displaying the corresponding blood vessel positions mapped to the blood vessel 3D model according to the display size information corresponding to the pre-stored level information. For example, a certain plaque stenosis is divided into three levels of stenosis, each stenosis corresponds to a stenosis size, the trained model can identify the level of the plaque stenosis of the blood vessel, and the adapted stenosis size is displayed at the corresponding position.

If the human body is the human body, respectively establishing a corresponding initial model of coronary artery plaque parameters, an initial model of lower limb arterial plaque parameters, an initial model of cerebral artery plaque parameters and an initial model of renal artery plaque parameters for different vascular areas;

training the initial models, training plaque parameters including plaque stenosis degree, blood vessel wall thickness and plaque thickness, dividing the plaque parameters of a sample level into a plurality of levels according to the plaque stenosis degree, dividing the blood vessel wall thickness into a plurality of levels, and dividing the plaque thickness into a plurality of levels for indexing and training;

carrying out plaque parameter model identification on the frame images of the target blood vessel region to obtain the level information of plaque parameters at the corresponding positions of the blood vessel;

and displaying the corresponding blood vessel positions mapped to the blood vessel 3D model according to the display size information corresponding to the pre-stored level information.

The mode can reflect the human twin blood vessel corresponding to the real human body. Considering the material, the method may further include:

establishing a blood vessel material model including wall elasticity and hardness;

training blood vessel material models of different target areas in advance;

carrying out blood vessel material model identification on the frame images of the target blood vessel region to obtain the material information of the blood vessel;

and displaying the corresponding blood vessel materials mapped to the blood vessel 3D model according to the pre-stored blood vessel material display type information.

The vessel material model identification may further comprise:

1) The arterial blood vessel of human body is composed of three layers of structures of an inner membrane, a middle membrane and an outer membrane. The thickness of each layer is different for blood vessels of different calibers and sites. Wherein:

_· the intima is in direct contact with blood and is made up of endothelial cells attached to a connective tissue bed of the basement membrane. The endothelial cell layer prevents the activation of coagulation and complement factors and inhibits the adhesion of leukocytes and platelets. In addition, it is involved in the regulation of vasoconstrictor dilation, growth and vascular remodeling.

_· The tunica media is the middle layer of the vessel wall, consisting mainly of Smooth Muscle Cells (SMC), as well as layers of elastic tissue and small amounts of collagen. The intima layer helps the blood vessel resist repeated distension and contraction caused by physiological pulsations in the blood flow and intra-luminal pressure.

_· The adventitia layer is composed of loose connective tissue and is mainly composed of fibroblasts, and when a blood vessel is damaged, the fibroblasts have the ability to repair the adventitia [.

_· Of these three layers, the middle membrane SMC, collagen and elastin fibers mainly play a role in ensuring the mechanical strength and elasticity of the blood vessel.

The different body areas differ in the material of the corresponding blood vessels, in particular in the texture, shape and thickness.

The blood vessel material model mainly comprises blood vessel texture characteristics, shape characteristics and thickness characteristics. In this example, the vascular features include vascular texture features and vascular thickness features.

In this example, the vessel texture features are implemented by a gray level co-occurrence matrix.

First, the frame images of the target blood vessel region are converted into grayscale images, and then, one direction (vertical in this example) and a separation distance D between pixels are selected to calculate a co-occurrence matrix.

For each pixel, the gray level co-occurrence frequency between neighboring pixels in the vertical direction at a distance D is counted. A gray level co-occurrence matrix is constructed, typically having a size of N x N, where N is the number of gray levels of the image. The element GLCM (i, j) in the co-occurrence matrix represents the number of simultaneous occurrences of pixel gray levels i and j at a given direction, distance.

Then, the element GLCM_prob (i, j) in the target vessel region co-occurrence probability matrix is obtained

Again, from the co-occurrence probability matrix, some statistics for blood line texture characterization, such as blood line contrast, correlation, energy, entropy, etc., may be calculated. These statistics can be used to further analyze the texture features of the blood line image.

2) Labeling a sample: and manually labeling the collected blood vessel image samples, namely labeling each sample with a corresponding blood vessel material characteristic value or class label. These labels can be defined according to practical situations, such as the density of blood vessel materials, texture roughness, etc.

Model selection and training: an appropriate machine learning or deep learning model is selected to train the vessel material model. Common models include Support Vector Machines (SVM), random forests (Random

Forest), convolutional Neural Network (CNN), and the like. And dividing the data into a training set and a verification set according to the result of the feature extraction and the labeling sample, and training a model by using the training set. And carrying out blood vessel material model identification on the frame images of the target blood vessel region to obtain the material information of the blood vessel.

Referring to fig. 2, a human vascular digital twin system 400 based on real world research includes a processing device 410 and a display device 430, the processing device 410 being further configured to:

acquiring medical images of a target blood vessel region, dividing image frames according to an image frame time sequence, and automatically extracting the blood vessels from the frame images through a blood vessel dividing model;

performing vessel 3D modeling according to the vessel attribute data;

dividing the medical image into a plurality of blocks, dividing each block into a plurality of blood vessel video sequences, dividing each blood vessel video sequence into video blocks at equal intervals, obtaining the blood flow velocity of each blood vessel of the video blocks according to the time sequence association among frames in the video blocks, and mapping the blood flow velocity into corresponding blood vessels in a blood vessel 3D model so as to construct the blood flow information of each blood vessel in the blood vessel 3D model corresponding to the medical image;

carrying out vascular plaque parameter model identification on the frame images of the target vascular region so as to map the images into corresponding blood vessels of the vascular 3D model;

and the display device 430 is used for displaying the blood vessel information of the blood vessel 3D model.

Generally, a memory 420 is also included that holds the various models.

When the various functions are integrated in an AI-like device, such as a wearable glasses, the present example may provide only one human vascular digital twinning device based on real world research, with the AI device configured to: acquiring medical images of a target blood vessel region, dividing image frames according to an image frame time sequence, and automatically extracting the blood vessels from the frame images through a blood vessel dividing model;

performing vessel 3D modeling according to the vessel attribute data;

and carrying out vascular plaque parameter model identification on the frame images of the target vascular region so as to map the frame images into corresponding blood vessels of a vascular 3D model, and displaying vascular information of the vascular 3D model through AI display.

For example, after the blood vessel information of the blood vessel 3D model can be displayed on the corresponding portion, the blood vessel information can be mapped to the human body 3D contour information, the preliminary human body contour is divided into a plurality of grids, the grid position can be recorded in the area of the target position to be detected, for example, a certain blood vessel is used as the target of detection, the network position is the network area 2 (for example, the network related to the head of the person is 1, the grid related to the face is 2, and the grid related to the neck is 3 …), and the network position appears on the third image. The mapping is performed, and some correction processing can be performed.

In one embodiment, a readable storage medium is provided, where the computer readable instructions, when executed by one or more processors, cause the one or more processors to perform the above steps, and specific steps are not described herein.

It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or partly in the prior art or in whole or in part in the form of a software product, which is stored in a storage medium, comprising instructions for causing a computer device (which may be a personal computer,

a server, or a network device, etc.) performs all or part of the steps of the methods described in the various embodiments of the invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a random access memory (random access memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus/network device and method may be implemented in other manners. For example, the apparatus/network device embodiments described above are merely illustrative, e.g., the division of the modules or units is merely a logical functional division, and there may be additional divisions in actual implementation, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection via interfaces, devices or units, which may be in electrical, mechanical or other forms.

The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

The above embodiments are only for illustrating the technical solution of the present application, and are not limiting; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application, and are intended to be included in the scope of the present application.

Claims

1. A human vascular digital twinning method based on real world research, comprising:

performing vessel 3D modeling according to the vessel attribute data;

and carrying out vascular plaque parameter model identification on the frame images of the target vascular region so as to map the images into corresponding blood vessels of the vascular 3D model.

2. The system of claim 1, wherein the automatic vessel extraction further comprises:

3. The system of claim 2, wherein vessel 3D modeling further comprises:

To obtain a 3D model of the blood vessel.

4. The method of claim 1, wherein dividing the medical image into a plurality of blocks, each block being further divided into a plurality of vessel video sequences, dividing each vessel video sequence into video blocks at equal intervals, training a corresponding neural network model, and obtaining blood flow velocities of vessels of the video block based on temporal correlations between frames in the video block further comprises:

5. The method of claim 1, wherein constructing blood flow information for each vessel in the 3D model of vessels corresponding to the medical image further comprises:

6. The method of claim 1, wherein,

firstly, respectively establishing a corresponding initial model of coronary artery plaque parameters, an initial model of lower limb arterial plaque parameters, an initial model of cerebral artery plaque parameters and an initial model of renal artery plaque parameters for different vascular areas;

7. The method as recited in claim 1, further comprising:

training blood vessel material models of different target areas in advance;

8. A human vascular digital twinning system based on real world research, comprising a processing device and a display device, the processing device further configured to:

performing vessel 3D modeling according to the vessel attribute data;

and the display device is used for displaying the blood vessel information of the blood vessel 3D model.

9. A human vascular digital twin device based on real world research, characterized by AI devices configured to: acquiring medical images of a target blood vessel region, dividing image frames according to an image frame time sequence, and automatically extracting the blood vessels from the frame images through a blood vessel dividing model;

performing vessel 3D modeling according to the vessel attribute data;