Non-invasive ultrasonic blood viscosity imaging methods, devices, equipment, media and products

Abstract

The present application discloses a non-invasive ultrasonic blood viscosity imaging method, apparatus, device, medium, and product, relating to the field of blood viscosity technology. The method uses a cross-beam-based vector Doppler imaging method to acquire a two-dimensional blood flow velocity vector of the blood vessel to be measured; establishes an incompressible Navier-Stokes equation for describing blood flow and encodes it as a loss function; combines the loss function with a physical information neural network to construct a blood viscosity measurement model; the blood viscosity measurement model includes a first physical information neural network and a second physical information neural network; the two-dimensional spatial coordinates and acquisition time of the two-dimensional blood flow velocity vector are used as the first input data, the two-dimensional spatial coordinates are used as the second input data, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector is used as a label. These are input into the blood viscosity measurement model together, and the blood viscosity is output to form a blood viscosity image of the blood vessel to be measured. The present application can non-invasively image the blood viscosity of the blood vessel.

Description

Technical Field

[0001] The present application relates to the technical field of blood viscosity, and in particular to a non-invasive ultrasonic blood viscosity imaging method, device, equipment, medium and product. Background Art

[0002] Blood viscosity is an important indicator reflecting the rheological properties of blood. Its essence is the internal friction resistance generated by the relative displacement of adjacent fluid layers when blood flows. Clinically, blood viscosity measurement can provide an auxiliary basis for the diagnosis of various diseases and has important reference value for risk assessment, treatment monitoring and prevention strategy formulation. In cases of dehydration, diabetes, atherosclerosis, etc., blood viscosity is high, which may lead to increased blood flow resistance, increased heart burden, and even promote the formation of blood clots. In certain diseases or drug treatments, blood viscosity may decrease, causing excessive blood flow or platelet activation in blood vessels, increasing the risk of bleeding.

[0003] Blood viscosity varies in distribution within blood vessels, exhibiting different distribution values ​​at different vascular locations and under different blood flow conditions. Currently, there are two main methods for measuring blood viscosity: capillary tube method and rotational method. The capillary tube method is the earliest technique used for measuring blood viscosity. Its basic principle is based on Poiseuille's law. Poiseuille's law describes the relationship between the flow rate of a fluid flowing steadily through a pipe and its viscosity, pipe radius, pipe length, and pressure differential. Poiseuille's law states that the flow rate is directly proportional to the pressure differential across the pipe, directly proportional to the fourth power of the pipe radius, and inversely proportional to the pipe length and the fluid viscosity. Therefore, the flow rate or flow rate of blood through a capillary tube can be used to calculate blood viscosity. The rotational method is a later-developed blood viscosity measurement method that uses the definition of viscosity as the ratio of shear stress to shear rate (or shear rate). In the rotational method, the sample is placed between a rotating disk or cylinder. The rotational motion generates shear forces, and viscosity is calculated by measuring the generated stress and rotation rate. In a rotational viscometer, the relationship between the rotation of a rotating surface (usually a disk or cylinder) and the resistance of a liquid is measured. By measuring the friction generated by the liquid during rotation, the viscosity of the liquid can be inferred.

[0004] However, both of these measurement methods are invasive and require blood sampling. There are also some non-invasive measurement methods, such as using two air cuffs or collecting fingertip volume pulse waves to calculate the waveform coefficient and deduce blood viscosity.

[0005] In summary, the current blood viscosity measurement method can only measure a blood viscosity value, but cannot image the blood viscosity and cannot intuitively display the two-dimensional distribution of blood viscosity. Summary of the Invention

[0006] The purpose of this application is to provide a non-invasive ultrasonic blood viscosity imaging method, device, equipment, medium and product, which can image the blood viscosity of blood vessels non-invasively.

[0007] To achieve the above objectives, this application provides the following solutions.

[0008] In a first aspect, the present application provides a non-invasive ultrasonic blood viscosity imaging method, comprising:

[0009] Using a cross-beam vector Doppler imaging method, the two-dimensional blood flow velocity vector of the blood vessel to be measured is collected within a preset time period;

[0010] Establish the incompressible Navier-Stokes equations for describing blood flow;

[0011] Encoding the incompressible Navier-Stokes equations as a loss function;

[0012] Combining the loss function and the physical information neural network, a blood viscosity measurement model is constructed; the blood viscosity measurement model includes a first physical information neural network and a second physical information neural network; the first physical information neural network is used to output a two-dimensional blood flow velocity based on the two-dimensional spatial coordinates and the acquisition time; the second physical information neural network is used to output the blood viscosity based on the two-dimensional spatial coordinates; the loss function is used to calculate the loss based on the two-dimensional blood flow velocity output by the first physical information neural network, the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector, and the blood viscosity output by the second physical information neural network;

[0013] The two-dimensional spatial coordinates in the two-dimensional blood flow velocity vector and the acquisition time of the two-dimensional blood flow velocity vector are used as the first input data, the two-dimensional spatial coordinates are used as the second input data, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector is used as a label. These are input together into the blood viscosity measurement model, the blood viscosity is output, and a blood viscosity image of the blood vessel to be measured is formed.

[0014] Optionally, the incompressible Navier-Stokes equations are:

[0015] ;

[0016] ;

[0017] ;

[0018] Where, is the fluid density, is the two-dimensional blood flow velocity vector field, is the collection time, is the gradient operator, is the pressure scalar field, is the unit tensor, is the viscous stress tensor, It is blood viscosity.

[0019] Optionally, the loss function is:

[0020] ;

[0021] ;

[0022] ;

[0023] ;

[0024] Where, is the total loss, It is data loss, is the first physical information loss, It is the second physical information loss; is the transverse component of the two-dimensional blood flow velocity vector, is the longitudinal component of the two-dimensional blood flow velocity vector, is the transverse blood flow velocity output by the first physical information neural network, is the longitudinal blood flow velocity output by the first physical information neural network, is the L2 norm; is the horizontal coordinate of the two-dimensional space coordinate, It is the vertical coordinate of the two-dimensional space coordinate.

[0025] Optionally, the input data dimension of the first physical information neural network is 3, the output data dimension is 2, the number of hidden layers is 9, and the number of neurons in each hidden layer is 80; the input data dimension of the second physical information neural network is 2, the output data dimension is 1, the number of hidden layers is 5, and the number of neurons in each hidden layer is 40.

[0026] Optionally, the two-dimensional spatial coordinates of the two-dimensional blood flow velocity vector and the acquisition time of the two-dimensional blood flow velocity vector are used as first input data, the two-dimensional spatial coordinates are used as second input data, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector is used as a label, and are input into the blood viscosity measurement model together to output the blood viscosity, specifically including:

[0027] Selecting the two-dimensional blood flow velocity vector used in this round from the collected two-dimensional blood flow velocity vectors of the blood vessels to be measured;

[0028] The two-dimensional spatial coordinates and acquisition time of the two-dimensional blood flow velocity vector used in this round are used as the first input data of this round, the two-dimensional spatial coordinates are used as the second input data of this round, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector is used as the label of this round, and are input into the blood viscosity measurement model together;

[0029] The first physical information neural network receives first input data and outputs a pressure scalar field and an intermediate physical field according to the first input data;

[0030] Calculate the partial differential of the intermediate physical field in two-dimensional space and output two-dimensional blood flow velocity;

[0031] The second physical information neural network receives the second input data and outputs the blood viscosity according to the second input data;

[0032] calculating data loss according to the output two-dimensional blood flow velocity and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector;

[0033] calculating a first physical information loss and a second physical information loss according to the output two-dimensional blood flow velocity, the output pressure scalar field, and the output blood viscosity;

[0034] Obtain the total loss of this round based on the data loss, the first physical information loss, and the second physical information loss;

[0035] If satisfied and , then the next round is carried out, and the step of selecting the two-dimensional blood flow velocity vector used in this round from the two-dimensional blood flow velocity vectors of the blood vessels to be measured is returned to be executed; wherein, is the total loss of this round, is the initial total loss, is the learning rate, is the number of iterations, is the maximum number of iterations;

[0036] If satisfied or , the blood viscosity output in this round is taken as the final blood viscosity.

[0037] Optionally, calculating the partial differential of the intermediate physical field in two-dimensional space and outputting the two-dimensional blood flow velocity specifically includes: according to the intermediate physical field, using the formula and , obtain the two-dimensional blood flow velocity; where, is the intermediate physical field, is the horizontal coordinate of the two-dimensional space coordinate, It is the vertical coordinate of the two-dimensional space coordinate.

[0038] In a second aspect, the present application provides a non-invasive ultrasonic blood viscosity imaging device, comprising: an acquisition module, an equation building module, an encoding module, a model building module and an application module.

[0039] The acquisition module is used to acquire the two-dimensional blood flow velocity vector of the blood vessel to be measured within a preset time period using a cross-beam based vector Doppler imaging method.

[0040] Equation building module for building the incompressible Navier-Stokes equations used to describe blood flow.

[0041] An encoding module for encoding the incompressible Navier-Stokes equations into a loss function.

[0042] A model construction module is used to combine the loss function and the physical information neural network to construct a blood viscosity measurement model; the blood viscosity measurement model includes a first physical information neural network and a second physical information neural network; the first physical information neural network is used to output a two-dimensional blood flow velocity based on the two-dimensional spatial coordinates and the acquisition time; the second physical information neural network is used to output the blood viscosity based on the two-dimensional spatial coordinates; the loss function is used to calculate the loss based on the two-dimensional blood flow velocity output by the first physical information neural network, the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector, and the blood viscosity output by the second physical information neural network.

[0043] The application module is used to input the two-dimensional spatial coordinates in the two-dimensional blood flow velocity vector and the acquisition time of the two-dimensional blood flow velocity vector as first input data, the two-dimensional spatial coordinates as second input data, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector as a label, input them together into the blood viscosity measurement model, output the blood viscosity, and form a blood viscosity image of the blood vessel to be measured.

[0044] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the above-described non-invasive ultrasonic blood viscosity imaging methods.

[0045] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned non-invasive ultrasonic blood viscosity imaging methods.

[0046] In a fifth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements any one of the above-described non-invasive ultrasonic blood viscosity imaging methods.

[0047] According to the specific embodiments provided in this application, this application has the following technical effects.

[0048] The present application provides a non-invasive ultrasonic blood viscosity imaging method, apparatus, equipment, medium and product, which uses a cross-beam based vector Doppler imaging method to collect two-dimensional blood flow velocity vectors of the blood vessels to be measured within a preset time period. The vector Doppler imaging method is an ultrasonic blood flow velocity measurement method that can non-invasively collect two-dimensional vector blood flow data; the incompressible Navier-Stokes equations are encoded into a loss function of a physical information neural network, thereby gradually converging to the blood viscosity of a real blood vessel that meets the conditions through the physical information neural network, thereby realizing non-invasive imaging of the blood viscosity of the blood vessels. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0050] Figure 1 Schematic diagram of a flow chart of a non-invasive ultrasonic blood viscosity imaging method in one embodiment of the present application.

[0051] Figure 2 A schematic diagram of the workflow of a blood viscosity measurement model provided in another embodiment of the present application.

[0052] Figure 3 A schematic diagram comparing the method of the present application and the existing method provided in another embodiment of the present application.

[0053] Figure 4 A schematic diagram of the functional modules of a non-invasive ultrasonic blood viscosity imaging device provided in one embodiment of the present application.

[0054] Figure 5 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0056] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0057] In an exemplary embodiment, Figure 1 As shown, a non-invasive ultrasonic blood viscosity imaging method is provided, including the following steps 101 to 105.

[0058] Step 101: Using a cross-beam based vector Doppler imaging method, a two-dimensional blood flow velocity vector of a blood vessel to be measured is collected within a preset time period.

[0059] Step 102: Establish the incompressible Navier-Stokes equations for describing blood flow.

[0060] Step 103: Encode the incompressible Navier-Stokes equations into a loss function.

[0061] Step 104: Construct a blood viscosity measurement model by combining the loss function and the physical information neural network; the blood viscosity measurement model includes a first physical information neural network and a second physical information neural network; the first physical information neural network is used to output a two-dimensional blood flow velocity according to the two-dimensional spatial coordinates and the acquisition time; the second physical information neural network is used to output the blood viscosity according to the two-dimensional spatial coordinates; the loss function is used to calculate the loss based on the two-dimensional blood flow velocity output by the first physical information neural network, the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector, and the blood viscosity output by the second physical information neural network.

[0062] Step 105: The two-dimensional spatial coordinates in the two-dimensional blood flow velocity vector and the acquisition time of the two-dimensional blood flow velocity vector are used as the first input data, the two-dimensional spatial coordinates are used as the second input data, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector is used as the label, and are input together into the blood viscosity measurement model, the blood viscosity is output, and a blood viscosity image of the blood vessel to be measured is formed.

[0063] By implementing the above steps 101 to 105, the cross-beam based vector Doppler imaging method uses ultrasonic vector flow imaging technology. Therefore, the present application realizes non-invasive imaging of blood viscosity of human blood vessels based on Physics-Informed Neural Networks (PINN) and the above ultrasonic vector flow imaging technology.

[0064] In another exemplary embodiment of the present application, the above step 101 uses a portable ultrasound device to place the probe on the blood vessel to be measured of the object to be examined (e.g., the human body), and uses cross-beam-based vector Doppler imaging technology to collect its two-dimensional vector blood flow data. Vector Doppler imaging technology is an advanced ultrasound blood flow velocity measurement method that can simultaneously obtain the transverse blood flow velocity ( direction) and longitudinal blood flow velocity ( direction), to achieve accurate reconstruction of the two-dimensional velocity vector. Two sets of blood flow velocity data can be obtained by ultrasound equipment collection, namely and , Represents the horizontal blood flow velocity component within a preset time period, Represents the longitudinal blood flow velocity component within a preset time period. These two sets of data will be used to input the neural network for blood viscosity imaging.

[0065] In another exemplary embodiment of the present application, the flow of blood can be approximately described by the incompressible Navier-Stokes (NS) equations. The incompressible Navier-Stokes equations are:

[0066] ;

[0067] ;

[0068] ;

[0069] Where, is the fluid density, is the two-dimensional blood flow velocity vector field, is the collection time, is the gradient operator, is the pressure scalar field, is the unit tensor, is the viscous stress tensor, It is blood viscosity.

[0070] In another exemplary embodiment of the present application, in order to better understand and predict the behavior of blood flow, it is usually necessary to use numerical methods to solve the incompressible Navier-Stokes equations. Common methods include finite difference method, finite element method and spectral method. These numerical methods can help simulate the flow of blood in complex vascular networks, including factors such as the curvature, branching and changes of blood vessels, and are of great significance for diagnosing vascular diseases (such as arteriosclerosis, hypertension, etc.) and studying hemodynamics. In this application, a physical information neural network is used to solve the above-mentioned incompressible Navier-Stokes equations, that is, the above-mentioned NS equations are encoded as the loss function of PINN, so as to image the blood viscosity.

[0071] The loss function consists of two parts. The first part is The purpose is to mark the actual observation data (that is, the two sets of data collected as and ), and the corresponding neural network prediction output and Minimize losses; the second part is , encoding the governing equations of the physical problem (the aforementioned NS equations) as part of the neural network training.

[0072] The loss function is expressed as follows based on the incompressible Navier-Stokes equations:

[0073] ;

[0074] ;

[0075] ;

[0076] ;

[0077] Where, is the total loss, It's data loss. is the first physical information loss, It is the second physical information loss; is the transverse component of the two-dimensional blood flow velocity vector, is the longitudinal component of the two-dimensional blood flow velocity vector, is the transverse blood flow velocity output by the first physical information neural network, is the longitudinal blood flow velocity output by the first physical information neural network, is the L2 norm; is the horizontal coordinate of the two-dimensional space coordinate, It is the vertical coordinate of the two-dimensional space coordinate.

[0078] In another exemplary embodiment of the present application, a physical information neural network is an innovative computational method that directly incorporates physical laws into the training process of the neural network by encoding the governing equations (partial differential equations) of the physical problem into a fully connected neural network. This method not only utilizes the powerful data learning and pattern recognition capabilities of deep learning models, but also effectively combines the laws of physics, making the model more robust when dealing with complex problems and having better physical interpretability. With the deepening of research, PINN has gradually become a powerful tool for solving direct and inverse problems, especially when facing complex problems such as multi-physical field coupling, showing its huge application potential and broad prospects.

[0079] The first physical information neural network of this application has an input data dimension of 3, an output data dimension of 2, a number of hidden layers of 9, and a number of neurons in each hidden layer of 80. The second physical information neural network has an input data dimension of 2, an output data dimension of 1, a number of hidden layers of 5, and a number of neurons in each hidden layer of 40. Specific network key parameters are shown in Table 1.

[0080] Table 1 Key network parameters

[0081]

[0082] Exemplarily, the activation function in this application may also use other activation functions, such as ReLu. During the model training process, only the weights of the neurons are changed, and the activation function does not change during the training process. This application may also use a B-spline function, the parameters of which can be updated during the training process, so that the activation function corresponding to each neuron will continue to change with the training process, thereby solving the partial differential equation.

[0083] The blood viscosity measurement model can also use network key parameters different from those given in Table 1, for example: the network structure has one more or one less layer, the number of neurons has 10 more or 5 less, etc.

[0084] The workflow of the blood viscosity measurement model designed in this application is as follows Figure 2 shown.

[0085] The first physical information neural network:

[0086] The input is , representing discrete space coordinates and discrete time series respectively. Output: intermediate physical field right Taking the derivative and multiplying by -1 we get ,right Taking the derivative we get , such a setting allows the network to automatically meet the incompressible condition ,because , ,So , another output is the pressure scalar field.

[0087] Second physical information neural network:

[0088] The input is , represent discrete space coordinates, there is no This is because blood viscosity hardly changes in a short period of time. Output is the blood viscosity to be obtained. The loss function of the second physical information neural network consists of two parts: and .

[0089] The process of iterative training of the blood viscosity measurement model is as follows.

[0090] To measure a section of the blood vessel plane, there are discrete two-dimensional spatial coordinates , measured for a period of time, and the acquisition time was obtained , so the Input the first physical information neural network, the predicted output of the first physical information neural network and First, the collected data (marked as and ) to minimize data loss, that is, the formula .

[0091] Since blood viscosity hardly changes in a short time, only the discrete two-dimensional space coordinates Input into the second physical information neural network, the model will predict the output blood viscosity .

[0092] because , the prediction output of the first physical information neural network , and , blood viscosity output by the second physical information neural network is satisfied with the incompressible Navier-Stokes equations mentioned above, so substitute these variables into and , to minimize physical information loss.

[0093] Combining the three loss terms, the total loss is The network loss is then propagated back, and the network parameters are iteratively optimized using the Adam optimizer (learning rate lr = 2e-4). Alternatively, the network optimizer can use other optimizers, such as Stochastic Gradient Descent (SGD).

[0094] In the iterative process, according to the dual constraints of the collected data and the physical equations, the blood viscosity output by the second physical information neural network is , from the initial chaotic and meaningless values, it will gradually converge to the blood viscosity of the real blood vessels that meet the conditions. The second physical information neural network will output a corresponding blood viscosity numerical value, thus achieving non-invasive blood viscosity imaging of human blood vessels.

[0095] Therefore, the above step 105 can be replaced by the following steps 201 to 210.

[0096] Step 201: Selecting a two-dimensional blood flow velocity vector used in this round from the collected two-dimensional blood flow velocity vectors of the blood vessels to be measured.

[0097] Step 202: The two-dimensional spatial coordinates and acquisition time of the two-dimensional blood flow velocity vector used in this round are used as the first input data of this round, the two-dimensional spatial coordinates are used as the second input data of this round, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector is used as the label of this round, and are input into the blood viscosity measurement model together.

[0098] Step 203: The first physical information neural network receives the first input data and outputs the pressure scalar field and the intermediate physical field according to the first input data.

[0099] Step 204: Calculate the partial differential of the intermediate physical field in two-dimensional space and output the two-dimensional blood flow velocity.

[0100] For example, according to the intermediate physical field, the formula and , and obtain two-dimensional blood flow velocity.

[0101] Step 205: The second physical information neural network receives the second input data and outputs blood viscosity according to the second input data.

[0102] Step 206: Calculate data loss based on the output two-dimensional blood flow velocity and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector.

[0103] Step 207: Calculate the first physical information loss and the second physical information loss according to the output two-dimensional blood flow velocity, the output pressure scalar field, and the output blood viscosity.

[0104] Step 208: Obtain the total loss of this round according to the data loss, the first physical information loss, and the second physical information loss.

[0105] Step 209: If satisfied and , then the next round is carried out, and the step of selecting the two-dimensional blood flow velocity vector used in this round from the two-dimensional blood flow velocity vectors of the blood vessels to be measured is returned to be executed; wherein, is the total loss of this round, is the initial total loss, is the learning rate, is the number of iterations, is the maximum number of iterations.

[0106] Step 210: If satisfied or , the blood viscosity output in this round is taken as the final blood viscosity.

[0107] The comparison between the method of the present application and the existing method is as follows: Figure 3As shown, both existing invasive and non-invasive measurement methods cannot image blood viscosity, while the method of the present application can image the blood viscosity of blood vessels non-invasively.

[0108] The advantages of the method of this application are: (1) it images blood viscosity rather than just a numerical value. (2) it is non-invasive. (3) it uses a radiation-free, highly safe ultrasound device. (4) the solution process is based on the NS equations, encoding the physical equations as the loss function of the neural network, so that the network does not learn data features in a disorderly manner, but instead uses physical laws as a guide to achieve parameter solution.

[0109] Based on the same inventive concept, embodiments of the present application also provide a non-invasive ultrasonic blood viscosity imaging device for implementing the aforementioned non-invasive ultrasonic blood viscosity imaging method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more embodiments of the non-invasive ultrasonic blood viscosity imaging device provided below can be found in the above-described limitations of the non-invasive ultrasonic blood viscosity imaging method and will not be further elaborated here.

[0110] In an exemplary embodiment, Figure 4 As shown, a video tag processing device is provided, which includes: an acquisition module, an equation building module, an encoding module, a model building module and an application module.

[0111] An acquisition module is used to acquire the two-dimensional blood flow velocity vector of the blood vessel to be measured within a preset time period using a cross-beam based vector Doppler imaging method; an equation establishment module is used to establish the incompressible Navier-Stokes equations for describing blood flow; and an encoding module is used to encode the incompressible Navier-Stokes equations into a loss function.

[0112] A model construction module is used to combine the loss function and the physical information neural network to construct a blood viscosity measurement model; the blood viscosity measurement model includes a first physical information neural network and a second physical information neural network; the first physical information neural network is used to output a two-dimensional blood flow velocity based on the two-dimensional spatial coordinates and the acquisition time; the second physical information neural network is used to output the blood viscosity based on the two-dimensional spatial coordinates; the loss function is used to calculate the loss based on the two-dimensional blood flow velocity output by the first physical information neural network, the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector, and the blood viscosity output by the second physical information neural network.

[0113] The application module is used to input the two-dimensional spatial coordinates in the two-dimensional blood flow velocity vector and the acquisition time of the two-dimensional blood flow velocity vector as first input data, the two-dimensional spatial coordinates as second input data, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector as a label, input them together into the blood viscosity measurement model, output the blood viscosity, and form a blood viscosity image of the blood vessel to be measured.

[0114] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store blood viscosity images. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a non-invasive ultrasonic blood viscosity imaging method is implemented.

[0115] Those skilled in the art will understand that Figure 5 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present application and does not constitute a limitation on the computer device to which the solution of the present application is applied. A specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps of the above-mentioned method embodiments when executing the computer program.

[0116] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0117] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0118] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0119] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0120] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

[0121] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0122] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A non-invasive ultrasonic blood viscosity imaging method, characterized in that: include: Using a cross-beam vector Doppler imaging method, the two-dimensional blood flow velocity vector of the blood vessel to be measured is collected within a preset time period; Establish the incompressible Navier-Stokes equations for describing blood flow; Encoding the incompressible Navier-Stokes equations as a loss function; Combining the loss function and the physical information neural network, a blood viscosity measurement model is constructed; the blood viscosity measurement model includes a first physical information neural network and a second physical information neural network; the first physical information neural network is used to output a two-dimensional blood flow velocity based on the two-dimensional spatial coordinates and the acquisition time; the second physical information neural network is used to output the blood viscosity based on the two-dimensional spatial coordinates; The loss function is used to calculate the loss based on the two-dimensional blood flow velocity output by the first physical information neural network, the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector, and the blood viscosity output by the second physical information neural network; The two-dimensional spatial coordinates of the two-dimensional blood flow velocity vector and the acquisition time of the two-dimensional blood flow velocity vector are used as first input data, the two-dimensional spatial coordinates are used as second input data, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector is used as a label. These are input into the blood viscosity measurement model, the blood viscosity is output, and a blood viscosity image of the blood vessel to be measured is formed; The incompressible Navier-Stokes equations are: ； ； ； Where, is the fluid density, is the two-dimensional blood flow velocity vector field, is the collection time, is the gradient operator, is the pressure scalar field, is the unit tensor, is the viscous stress tensor, is blood viscosity; The loss function is: ； ； ； ； Where, is the total loss, It's data loss. is the first physical information loss, It is the second physical information loss; is the transverse component of the two-dimensional blood flow velocity vector, is the longitudinal component of the two-dimensional blood flow velocity vector, is the transverse blood flow velocity output by the first physical information neural network, is the longitudinal blood flow velocity output by the first physical information neural network, is the L2 norm; is the horizontal coordinate of the two-dimensional space coordinate, It is the vertical coordinate of the two-dimensional space coordinate.

2. The non-invasive ultrasonic blood viscosity imaging method according to claim 1, characterized in that: The first physical information neural network has an input data dimension of 3, an output data dimension of 2, a number of hidden layers of 9, and a number of neurons in each hidden layer of 80; The input data dimension of the second physical information neural network is 2, the output data dimension is 1, the number of hidden layers is 5, and the number of neurons in each hidden layer is 40.

3. The non-invasive ultrasonic blood viscosity imaging method according to claim 1, characterized in that: The two-dimensional spatial coordinates of the two-dimensional blood flow velocity vector and the acquisition time of the two-dimensional blood flow velocity vector are used as first input data, the two-dimensional spatial coordinates are used as second input data, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector is used as a label. These two data are input together into the blood viscosity measurement model, and the blood viscosity is output, specifically including: Selecting the two-dimensional blood flow velocity vector used in this round from the collected two-dimensional blood flow velocity vectors of the blood vessels to be measured; The two-dimensional spatial coordinates and acquisition time of the two-dimensional blood flow velocity vector used in this round are used as the first input data of this round, the two-dimensional spatial coordinates are used as the second input data of this round, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector is used as the label of this round, and are input into the blood viscosity measurement model together; The first physical information neural network receives first input data and outputs a pressure scalar field and an intermediate physical field according to the first input data; Calculate the partial differential of the intermediate physical field in two-dimensional space and output two-dimensional blood flow velocity; The second physical information neural network receives the second input data and outputs the blood viscosity according to the second input data; calculating data loss according to the output two-dimensional blood flow velocity and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector; calculating a first physical information loss and a second physical information loss according to the output two-dimensional blood flow velocity, the output pressure scalar field, and the output blood viscosity; Obtain the total loss of this round based on the data loss, the first physical information loss, and the second physical information loss; If satisfied and , then the next round is carried out, and the step of selecting the two-dimensional blood flow velocity vector used in this round from the two-dimensional blood flow velocity vectors of the blood vessels to be measured is returned to be executed; wherein, is the total loss of this round, is the initial total loss, is the learning rate, is the number of iterations, is the maximum number of iterations; If satisfied or , the blood viscosity output in this round is taken as the final blood viscosity.

4. The non-invasive ultrasonic blood viscosity imaging method according to claim 3, characterized in that: Calculate the partial differential of the intermediate physical field in two-dimensional space and output the two-dimensional blood flow velocity, including: According to the intermediate physical field, using the formula and , obtain the two-dimensional blood flow velocity; where, It is the intermediate physical field.

5. A non-invasive ultrasonic blood viscosity imaging device, characterized in that: include: An acquisition module, configured to acquire a two-dimensional blood flow velocity vector of a blood vessel to be measured within a preset time period using a cross-beam based vector Doppler imaging method; an equation building module for building the incompressible Navier-Stokes equations for describing blood flow; An encoding module for encoding the incompressible Navier-Stokes equations into a loss function; a model construction module, configured to construct a blood viscosity measurement model by combining the loss function and a physical information neural network; the blood viscosity measurement model comprising a first physical information neural network and a second physical information neural network; the first physical information neural network being configured to output a two-dimensional blood flow velocity based on two-dimensional spatial coordinates and acquisition time; and the second physical information neural network being configured to output blood viscosity based on the two-dimensional spatial coordinates; The loss function is used to calculate the loss based on the two-dimensional blood flow velocity output by the first physical information neural network, the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector, and the blood viscosity output by the second physical information neural network; an application module, configured to input the two-dimensional spatial coordinates of the two-dimensional blood flow velocity vector and the acquisition time of the two-dimensional blood flow velocity vector as first input data, the two-dimensional spatial coordinates as second input data, and the two-dimensional blood flow velocity in the two-dimensional blood flow velocity vector as a label, input them together into the blood viscosity measurement model, output the blood viscosity, and form a blood viscosity image of the blood vessel to be measured; The incompressible Navier-Stokes equations are: ； ； ； Where, is the fluid density, is the two-dimensional blood flow velocity vector field, is the collection time, is the gradient operator, is the pressure scalar field, is the unit tensor, is the viscous stress tensor, is blood viscosity; The loss function is: ； ； ； ； Where, is the total loss, It is data loss, is the first physical information loss, It is the second physical information loss; is the transverse component of the two-dimensional blood flow velocity vector, is the longitudinal component of the two-dimensional blood flow velocity vector, is the transverse blood flow velocity output by the first physical information neural network, is the longitudinal blood flow velocity output by the first physical information neural network, is the L2 norm; is the horizontal coordinate of the two-dimensional space coordinate, It is the vertical coordinate of the two-dimensional space coordinate.

6. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the non-invasive ultrasonic blood viscosity imaging method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the non-invasive ultrasonic blood viscosity imaging method according to any one of claims 1 to 4 is implemented.

8. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the non-invasive ultrasonic blood viscosity imaging method according to any one of claims 1 to 4 is implemented.

Images