A method, device and program product for calculating blood flow pressure

Abstract

The present application relates to the field of intelligent medical treatment, and specifically to a method, device and program product for calculating blood flow pressure. It includes S1: obtaining an acceleration-coded image of a patient's MRI; S2: inputting the acceleration-coded image into a pressure model to calculate the blood flow pressure; wherein the pressure model obtains the pixel-pressure relationship by calculating the amplitude and pressure gradient in the acceleration-coded image, and is constructed based on the pixel-pressure relationship. Compared with invasive catheter pressure measurement, the present application is non-invasive and risk-free, suitable for early diagnosis and auxiliary routine examinations, and has good clinical value.

Description

Technical Field

[0001] The present application relates to the field of intelligent healthcare, and particularly to a method, device, program product, and computer-readable storage medium for calculating blood flow pressure. Background Art

[0002] Blood flow pressure is an important reference index for the diagnosis of many cardiovascular diseases. In clinical examinations, the methods for measuring blood flow pressure are divided into invasive and non-invasive types. Invasive catheter pressure measurement is considered the gold standard for clinically measuring intracardiovascular pressure and is commonly used for auxiliary diagnosis of diseases and intensive care, etc. When assisting in the diagnosis of diseases, real-time scanning under X-ray is required to guide the catheter or wire into the target blood vessel or intracardiac position for direct pressure measurement. This process is invasive and has a certain degree of radiation damage. Therefore, the use of this technology in clinical practice is limited. Moreover, invasive catheter pressure measurement is usually only used in patients with severe symptoms, which may lead to delayed disease diagnosis and the inability to achieve early diagnosis and early treatment. Therefore, it is crucial to invent a non-invasive method for early measurement of blood flow pressure. With the development of non-invasive technologies, Doppler ultrasound and time-resolved four-dimensional phase-contrast magnetic resonance flow imaging technology (Four-Dimensional Magnetic Resonance Flow Imaging, 4D Flow MRI) have been introduced for non-invasive measurement of blood flow pressure. However, Doppler ultrasound can only estimate the pressure gradient at the site of vascular stenosis or valvular stenosis and cannot provide time and space pressure change information of the entire heart. Doppler echocardiography can obtain peak flow velocity information by imaging velocity without radiation conditions, and then calculate the relative pressure through the simplified Bernoulli equation. However, due to the low contrast of ultrasound images and the objective assumptions of the simplified Bernoulli equation, the accuracy of pressure measurement is limited. At the same time, the determination of the peak velocity mainly depends on the subjective experience of the ultrasound imaging operator, making the calculation result lack a certain degree of objectivity.

[0003] Compared with them, 4D Flow MRI shows great potential in evaluating pressure gradients in vivo with its non-invasive, non-radiative, and objective characteristics. Moreover, this technology also provides complete temporal and spatial fluid information such as three-dimensional spatial flow velocity or three-dimensional spatial acceleration. There are mainly two MRI methods: one is to calculate the pressure gradient based on the velocity vector map, by using spatial integration and iterative optimization to iteratively calculate the pressure distribution map of the pressure Poisson equation (PPE). However, solving the PPE requires boundary conditions, and the boundary conditions provided by 4D Flow MRI inevitably have noise and errors; the other is the pressure gradient calculation based on the acceleration vector map, which is initialized by arbitrarily setting the pressure at the vertex to zero in the time frames of each cardiac cycle. In the intraventricular pressure map calculated within the region of interest (ROI), the relative pressure changes throughout the cardiac cycle were first qualitatively evaluated. Then, along the manually drawn trajectory, the pressure changes of the blood flow across the mitral valve and across the aorta were measured in the frames of each cardiac cycle. This method is only limited to two-dimensional images, and setting the pressure at the vertex to zero affects the authenticity of the blood flow pressure measurement. Summary of the Invention

[0004] In order to achieve early non-invasive and accurate calculation of blood flow pressure, based on magnetic resonance 4D Flow imaging, the present invention proposes a method and system for calculating blood flow pressure, specifically including:

[0005] S1: Obtain the patient's MRI acceleration-encoded image;

[0006] S2: Input the acceleration-encoded image into a pressure model for calculation to obtain the blood flow pressure; wherein, the pressure model is constructed based on the pixel pressure relationship obtained by calculating the amplitude and pressure gradient in the acceleration-encoded image.

[0007] Furthermore, the construction method of the pressure model is as follows:

[0008] S101: Obtain the patient's MRI acceleration-encoded image;

[0009] S102: Obtain amplitude information based on the acceleration-encoded image to obtain an amplitude image;

[0010] S103: Obtain phase information based on the acceleration-encoded image to obtain a phase image;

[0011] S104: Obtain the pressure gradient of each pixel based on the pixel points of the phase image;

[0012] S105: Calculate the amplitude image and the pressure gradient to obtain a three-dimensional pressure gradient dataset;

[0013] S106: Obtain the pixel pressure relationship based on the three-dimensional pressure gradient dataset;

[0014] S107: Construct a pressure model based on the pixel pressure relationship;

[0015] Furthermore, the pixel pressure relationship is the relationship between the pixel pressure value and the pixel pressure gradient;

[0016] Optionally, the calculation in step S105 is a dot product calculation.

[0017] Furthermore, step S104 is specifically: first obtain the acceleration information of each pixel in the phase image, and then calculate the pressure gradient of each pixel based on the acceleration information of each pixel.

[0018] S102 further includes amplitude region delineation. An interested region is delineated from the amplitude image to obtain an amplitude region image, and the amplitude region image is calculated with the pressure gradient to obtain a three-dimensional data set;

[0019] Optionally, for the amplitude region delineation, first select the number of layers containing the interested region based on the cross-sectional image, and perform a first selection on the interested region of the number of layers to obtain a selected region image; perform a coronal plane reconstruction on the selected image to obtain a coronal plane image, and then perform a second selection on the coronal plane image to obtain an amplitude region image;

[0020] Optionally, the amplitude region image further includes background removal, where the pixels within the region in the amplitude region image are set to 1 and the pixels outside the region are set to 0.

[0021] The objective function E(P) of the pixel pressure relationship is expressed as:

[0022]

[0023] where, represents the pressure gradient between pixel points (i,j,k) and (i + 1,j,k); ψ i,j,k represents the pressure gradient between pixel points (i,j,k) and (i,j + 1,k); Θ i,j,k represents the pressure gradient between pixel points (i,j,k) and (i,j,k + 1), and P i,j,k represents the pressure value of pixel point (i,j,k), and M, N, and L respectively represent different dimensions.

[0024] The calculation is to calculate the blood flow pressure by a conjugate gradient algorithm for the pressure model;

[0025] Optionally, the conjugate gradient algorithm first takes the derivative of the objective function of the pressure model to obtain a gradient function, and the gradient function is iteratively calculated until the function converges to obtain the blood flow pressure;

[0026] Optionally, the iterative calculation is performed iteratively by updating the step size, solution vector, residual, and search direction;

[0027] Optionally, the step size determines the updated step size by calculating the objective function and gradient;

[0028] Optionally, the step size is updated by the following formula, expressed as:

[0029] α = r T r / p T Ap

[0030] where α represents the updated step size, r represents the residual vector of the current solution P, A represents the coefficient matrix, and p represents the current search direction vector;

[0031] Optionally, the direction is updated by the following formula, expressed as:

[0032] p = r + (r new T *r new / r old T *r old )p

[0033] where p represents the updated direction, r represents the residual vector of the current solution P, r new represents the sum of squares of the updated residual vectors, r old represents the sum of squares of the residual vectors of the previous iteration, r new T represents the transpose of r new represents the transpose of r old T represents the transpose of r old represents the transpose of r.

[0034] The obtaining of the patient's MRI acceleration-encoded image includes one or more of the following: 4D FLOW MRI, 3D FLOW MRI, 2D FLOW MRI;

[0035] Optionally, S1 is replaced with obtaining the patient's velocity-encoded image, and the acceleration-encoded image is calculated based on the velocity-encoded image;

[0036] Optionally, the acceleration-encoded image is obtained from the velocity-encoded image by differential calculation;

[0037] Optionally, the pressure calculation of the method further includes any one or more of the following: portal vein pressure, intracranial pressure.

[0038] The object of the present invention is to provide a computer program product having a computer program or instruction, and the computer program or instruction is executed by a processor to implement the above method for calculating blood flow pressure.

[0039] The object of the present invention is to provide a computer device, including a memory, a processor, and a computer program or instruction stored on the memory, including:

[0040] The processor executes the computer program or instruction to implement the method for calculating blood flow pressure described above.

[0041] The object of the present invention is to provide a computer-readable storage medium, on which a computer program or instruction is stored, and the computer program or instruction is executed by a processor to implement the method for calculating blood flow pressure described above.

[0042] Advantages of the present invention:

[0043] 1. An innovative mathematical model of pixel pressure value and pressure gradient between pixels is proposed to quickly calculate the value of blood flow pressure, which can provide time and space pressure change information of the whole heart, avoid the error caused by the simplified Bernoulli equation, and at the same time avoid the noise and error problems of the boundary conditions provided by 4D Flow MRI. Moreover, it is not limited to two-dimensional images and can obtain more dimensional information.

[0044] 2. Compared with invasive catheter pressure measurement, the method of the present invention is non-invasive and risk-free, and is suitable for early diagnosis and auxiliary routine examinations.

[0045] 3. Compared with the existing 4D Flow MRI calculation method based on velocity vector maps, the present invention uses three-dimensional acceleration data to avoid velocity calculation errors and improve calculation accuracy.

[0046] 4. An innovative mathematical model of pixel pressure value and pressure gradient between pixels is used to quickly calculate the value of blood flow pressure, and the calculation process is integrated into the interface developed by Matlab. Users only need to import acceleration data and delineate the region of interest to automatically complete the calculation of blood flow pressure.

[0047] 5. Compared with the existing 4D Flow MRI calculation method based on acceleration vector maps, the present invention does not need to assume boundary conditions and initialization parameters, has universality, and can calculate the pressure values of each pixel point during the whole cardiac cycle. Description of the Drawings

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative efforts.

[0049] Figure 1Schematic flowchart of a method for calculating blood flow pressure provided by an embodiment of the present invention;

[0050] Figure 2 Schematic diagram of a system for calculating blood flow pressure provided by an embodiment of the present invention;

[0051] Figure 3 Schematic diagram of a device for calculating blood flow pressure provided by an embodiment of the present invention;

[0052] Figure 4 Diagram of selecting a region of interest provided by an embodiment of the present invention. Detailed implementation manners

[0053] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0054] In some processes described in the specification and claims of the present invention and the above-mentioned drawings, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The serial numbers of the operations, such as S101, S102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0055] Figure 1 A schematic diagram for calculating blood flow pressure provided by an embodiment of the present invention specifically includes:

[0056] S1: Obtain the MRI acceleration-encoded image of the patient;

[0057] In one embodiment, the obtaining of the MRI acceleration-encoded image of the patient includes one or more of the following: 4DFLOW MRI, 3D FLOW MRI, 2D FLOW MRI.

[0058] In one embodiment, 4D Flow MRI, 3D Flow MRI, and 2D Flow MRI are different magnetic resonance imaging (MRI) techniques, which have their own characteristics and application scopes in hemodynamic research. These techniques mainly measure the velocity and direction of blood flow through phase contrast (PC) MRI.

[0059] 2D Flow MRI (Two-Dimensional Flow MRI) is the most basic phase-contrast MRI technique, which acquires blood flow information in a plane (usually the cross-section). This means that it can only measure the velocity vector of blood flow on a selected two-dimensional slice. 2D Flow MRI can provide information on the blood flow velocity and direction on a specific slice.

[0060] 3D Flow MRI (Three-Dimensional Flow MRI) is an upgraded technique that acquires blood flow information in three-dimensional space. Compared with 2D Flow MRI, 3D Flow MRI can simultaneously measure the blood flow velocities in three directions within a volume, thus providing a more comprehensive spatial blood flow distribution. However, 3D Flow MRI is usually acquired at fixed time points, so it lacks temporal resolution and cannot continuously observe the changes in blood flow over time.

[0061] 4D Flow MRI (Four-Dimensional Flow MRI) is a further development of 3D Flow MRI, which adds the time dimension on the basis of three-dimensional space and can dynamically monitor the changes in blood flow over time. 4D Flow MRI can provide blood flow velocity, direction, and hemodynamic parameters that change over time, such as vortices, shear stress, and pressure gradients. This technique allows doctors and researchers to not only observe the static blood flow distribution but also understand how hemodynamics change with the cardiac cycle, which is particularly important for understanding cardiovascular diseases. The main advantage of 4D Flow MRI is that it can "retrospectively" analyze blood flow, which means that even if the blood flow conditions change during the scan, the blood flow can be re-evaluated and measured during the data processing stage. In addition, 4D Flow MRI provides better spatial and temporal resolution and can capture more complex blood flow patterns, such as eddies and regurgitation, which are very valuable for clinical evaluation and scientific research analysis.

[0062] In one embodiment, velocity encoding is mainly applied in Phase Contrast MRI (PC-MRI) to quantify the flow rate of blood or other flowing substances. In PC-MRI, by applying a special pulse sequence, the motion velocity of the fluid can be encoded. This usually involves applying one or more velocity encoding (VENC) gradient fields, which introduce phase changes on the flowing protons, and the phase changes are proportional to the flow rate. By comparing the images with and without the VENC gradient field applied, the velocity and direction of the fluid can be calculated.

[0063] In one embodiment, S1 is replaced by obtaining a velocity-encoded image of the patient, and calculating an acceleration-encoded image based on the velocity-encoded image;

[0064] Optionally, the acceleration-encoded image is obtained by differential calculation from the velocity-encoded image.

[0065] In a specific embodiment, there are two methods for obtaining the acceleration-encoded image: (1) Scanning the patient using an MRI 4D Flow acceleration-encoded sequence, and the scanning result is the acceleration-encoded image;

[0066] (2) Scanning the patient using an MRI 4D Flow velocity-encoded sequence, and the scanning result is the velocity-encoded image. Other mathematical methods such as differential calculation method can be used to calculate the acceleration-encoded image from the velocity-encoded image.

[0067] S2: Input the acceleration-encoded image into a pressure model for calculation to obtain blood flow pressure; wherein, the pressure model is constructed based on the pixel pressure relationship obtained by calculating the amplitude and pressure gradient in the acceleration-encoded image.

[0068] In one embodiment, the Conjugate Gradient Method is an iterative algorithm mainly used to solve large-scale linear systems, especially those represented by symmetric positive definite matrices. The core idea of the Conjugate Gradient Method lies in finding a series of conjugate directions and searching along these directions to achieve fast convergence. Here, "conjugate" means that in a certain sense, these directions are orthogonal to each other. Specifically, they are orthogonal under the action of the matrix A (the coefficient matrix of the linear system). If A is symmetric positive definite, the inner product of the conjugate directions under A is zero.

[0069] In one embodiment, the construction method of the pressure model is as follows:

[0070] S101: Obtain the MRI acceleration-encoded image of the patient;

[0071] S102: Obtain amplitude information based on the acceleration-encoded image to obtain an amplitude image;

[0072] S103: Obtain phase information based on the acceleration-encoded image to obtain a phase image;

[0073] S104: Obtain the pressure gradient of each pixel based on the pixel points of the phase image;

[0074] S105: Calculate the amplitude image and the pressure gradient to obtain a three-dimensional dataset of the pressure gradient;

[0075] S106: Obtain the pixel pressure relationship based on the three-dimensional pressure gradient dataset;

[0076] S107: Construct a pressure model based on the pixel pressure relationship.

[0077] Optionally, the pixel pressure relationship is the relationship between the pixel pressure value and the pixel pressure gradient.

[0078] Optionally, the calculation in step S105 is a dot product calculation.

[0079] In one embodiment, step S104 is specifically: first obtain the acceleration information of each pixel in the phase image, and then calculate the pressure gradient of each pixel based on the acceleration information of each pixel.

[0080] In one embodiment, S102 further includes amplitude region delineation. An interested region is delineated from the amplitude image to obtain an amplitude region image, and the amplitude region image is calculated with the pressure gradient to obtain a three-dimensional dataset.

[0081] In one embodiment, the amplitude region delineation is to first select the number of layers containing the interested region based on the cross-sectional image, and perform a first selection on the interested region of the number of layers to obtain a selected region image; perform a coronal plane reconstruction on the selected image to obtain a coronal plane image, and then perform a second selection on the coronal plane image to obtain an amplitude region image.

[0082] In one embodiment, the amplitude region image further includes background removal, where the pixels within the region in the amplitude region image are set to 1 and the pixels outside the region are set to 0.

[0083] In one embodiment, the objective function E(P) of the pixel pressure relationship is expressed as:

[0084]

[0085] where, represents the pressure gradient between pixel points (i, j, k) and (i + 1, j, k); ψ i,j,k represents the pressure gradient between pixel points (i, j, k) and (i, j + 1, k); Θ i,j,k represents the pressure gradient between pixel points (i, j, k) and (i, j, k + 1), and P i,j,k represents the pressure value of pixel point (i, j, k), and M, N, and L respectively represent different dimensions.

[0086] In one embodiment, the calculation is to calculate the blood flow pressure of the pressure model through the conjugate gradient algorithm;

[0087] Optionally, the conjugate gradient algorithm first takes the derivative of the objective function of the pressure model to obtain a gradient function, and the gradient function is iteratively calculated until the function converges to obtain the blood flow pressure;

[0088] Optionally, the iterative calculation is performed by updating the step size, solution vector, residual, and search direction;

[0089] Optionally, the step size is determined by calculating the objective function and the gradient to update the step size;

[0090] Optionally, the step size is updated by the following formula, expressed as:

[0091] α = r T r / p T Ap

[0092] where α represents the updated step size, r represents the residual vector of the current solution P, A represents the coefficient matrix, and p represents the current search direction vector;

[0093] Optionally, the direction is updated by the following formula, expressed as:

[0094] p = r + (r new T *r new / r old T *r old )p

[0095] where p represents the updated direction, r represents the residual vector of the current solution P, r new represents the sum of the squares of the updated residual vectors, r old represents the sum of the squares of the residual vectors of the previous iteration, r new T represents the transpose of r new and r old T represents the transpose of r old represents the transpose of r

[0096] In one embodiment, the pressure calculation of the method further includes any one or more of the following: portal vein pressure, intracranial pressure.

[0097] In a specific embodiment, the image result after magnetic resonance scanning includes magnitude map information and phase map information. Generally speaking, the image obtained after scanning is equal to the magnitude map plus the phase map. In the calculation program of the present invention, in the first step, after reading the acceleration-encoded image, the magnitude information and phase information in the image are separately extracted to generate a magnitude map and a phase map. The magnitude map represents anatomical structure information, and the phase map represents motion information.

[0098] In a specific embodiment, the magnetic resonance acceleration encoded image is read and preprocessed. The magnetic resonance acceleration encoded image is first read into the blood flow pressure measurement software, which classifies the image into an amplitude image and a phase image. The amplitude image is used to delineate the region of interest in the next step, and the phase image is used to obtain the acceleration information of each pixel point. Based on the NS equation [1], it is written in the form of acceleration and can be expressed as formula [2].

[0099]

[0100] In conventional studies of blood flow pressure, the viscosity coefficient and external force terms are ignored, so the pressure gradient is simplified to formula [3]. Based on this, the pressure gradient of each pixel can be calculated through the magnetic resonance acceleration encoding image, which is used as a preparation for the third step of establishing a mathematical model.

[0101]

[0102] Furthermore, after reading the magnetic resonance acceleration encoded image, an amplitude image will be generated for ROI delineation. First, the number of layers containing the region of interest is selected based on the cross section. For example, taking the pulmonary artery as an example, the first ROI selection is performed after selecting the layer containing the pulmonary artery. The software will reconstruct the coronal plane of the pulmonary artery area, and then the second ROI selection can select the pulmonary artery area (such as Figure 4 shown). The software will set the ROI area in the entire three-dimensional space to 1 and the rest to 0, and then multiply it with the pressure gradient of each pixel to obtain the three-dimensional data set of the pressure gradient of the ROI area.

[0103] Furthermore, a mathematical model of pixel pressure value and pressure gradient between pixels, namely pressure model, is established. The objective function of the model is:

[0104]

[0105] Among them, represents the pressure gradient between pixel points (i,j,k) and (i+1,j,k); ψ i,j,k represents the pressure gradient between pixel points (i,j,k) and (i,j+1,k); Θ i,j,k represents the pressure gradient A between pixel points (i,j,k) and (i,j,k+1), and Pi,j,k represents the pressure value of pixel point (i,j,k). This model indicates that the difference between adjacent pixels should be close to the pressure gradient between them, and the pressure value of each pixel in each cardiac cycle is solved by minimizing the objective equation.

[0106] The first term in the above mathematical model: Represents the change of the pressure value P of each pixel in the X direction, Pi+1,j,k -P i,j,k represents the gradient of P in the X direction, φ i,j,k represents the gradient calculated based on the NS equation in Step 1. The sum of the squares of the differences between the actual gradient and the theoretical gradient of the pressure calculated by this term is minimized, indicating that the numerical value of the expected actual gradient approximates the theoretical gradient. The same applies to the second and third terms. The last term is the regularization coefficient, which plays a role in denoising and preventing overfitting of the results.

[0107] Furthermore, the conjugate gradient method is used to solve for the pressure. The conjugate gradient method is an effective numerical method for solving large linear equations Ax = b or optimizing quadratic functions. The following is a step-by-step description of the conjugate gradient method solution process in combination with the above code:

[0108] 1) Objective function E(P):

[0109]

[0110] 2) Gradient ▽E: The gradient gradE is obtained by taking the derivative of the objective function E(P) term by term;

[0111]

[0112] where, -2(P i+1,j,k -P i,j,k -Δ i,j,k ) + 2(P i,j,k -P i-1,j,k -Δ i-1,j,k ) represents in the x direction, -2(P i,j+1,k -P i,j,k -Γ i,j,k ) + 2(P i,j,k -P i,j-1,k -Γ i,j-1,k ) represents in the y direction, -2(P i,j,k +1 - P i,j,k -Λ i,j,k ) + 2(P i,j,k -P i,j,k -1 - Λ i,j,k-1 ) represents in the z direction, +2λP i,j,k represents the regularization term.

[0113] 3) Initial value setting:

[0114] P = P0(:), converting each pressure value of the three-dimensional pixel points into a column vector;

[0115] r = -gradE, using the negative gradient as the initial residual;

[0116] p = r, initializing the direction vector as the residual;

[0117] r sold = r T * r, calculate the update formula of the conjugate gradient method for the initial residual sum of squares, and assign r snew to r sold for use in the next iteration.

[0118] 4) Iterative calculation:

[0119] Step size: α = r T r / p T Ap. The step size is also the learning rate. By calculating the objective function and the gradient, ensure that the search direction and step size for each iteration are reasonable;

[0120] Update the solution vector: P = P + αP (update each pressure value);

[0121] Update the residual: r = r - αAp;

[0122] Update the search direction: p = r + (r new T * r new / r old T * r old )p. Update the search direction to ensure conjugate directions, thus accelerating convergence.

[0123] Among them, r (Residual): represents the residual vector of the current solution P, that is, r = b - A

[0124] * P, calculated in each iteration. The initial residual r is the negative gradient, indicating r = -gradE, because in the optimization problem, the residual and the gradient are negatively correlated.

[0125] r new (New Residual): represents the sum of squares of the updated residual vector, that is, r new =

[0126] r T * r, where r T is the transposed vector. After each update of the residual, calculate r snew = r T * r, representing the new sum of squares of the residual.

[0127] r old (Old Residual) represents the sum of squares of the residual vector in the previous iteration, that is, r sold =

[0128] r T * r. r soldUsed to save the sum of squared residuals from the previous iteration for calculating the step size and updating the search direction.

[0129] r T new and r T old : Here, r T new and r T old refers to the transpose of r new and r old . These symbols usually represent the transpose of a vector in mathematical derivations.

[0130] A (Matrix): Represents the coefficient matrix A, which is used to define the linear equation Ax = b in the conjugate gradient method. A is used to calculate Ap, that is, the product of A and the direction vector p.

[0131] p (Search Direction): Represents the current search direction vector. In each iteration, the search direction is updated according to the residual. The initial value of p is the residual r, and then it is updated to p = r + (r snew / r sold ) * p in each iteration.

[0132] Embodiments of the present disclosure also provide a computer program product or system, including a computer program, which when executed by a processor implements the steps of the above method for calculating blood flow pressure.

[0133] Figure 2 A schematic diagram of a system for calculating blood flow pressure provided by an embodiment of the present invention specifically includes:

[0134] An acquisition unit: acquires a patient's MRI acceleration-encoded image;

[0135] A calculation unit: inputs the acceleration-encoded image into a pressure model for calculation to obtain blood flow pressure; wherein, the pressure model is constructed based on the pixel pressure relationship obtained by calculating the amplitude and pressure gradient in the acceleration-encoded image.

[0136] Figure 3 A schematic diagram of a device for calculating blood flow pressure provided by an embodiment of the present invention specifically includes:

[0137] A memory and a processor; the memory is used to store program instructions; the processor is used to call the program instructions, and when the program instructions are executed, it performs any one of the above methods for calculating blood flow pressure.

[0138] An embodiment of the present disclosure also provides a computer-readable storage medium storing a computer program, which when executed by a processor, performs any one of the above-mentioned methods for calculating blood flow pressure.

[0139] The verification results of this verification embodiment show that allocating a fixed weight for an indication can improve the performance of this method compared to the default setting. Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein. In several embodiments provided by the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the couplings, direct couplings, or communication connections shown or discussed with each other can be indirect couplings or communication connections through some interfaces, devices, or units, and can be in electrical, mechanical, or other forms. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in each embodiment of the present invention, the functional units can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units. Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. The storage medium can include: read-only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), magnetic disk, or optical disk, etc.

[0140] Those of ordinary skill in the art can understand that all or part of the steps in implementing the methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. The above-mentioned medium storage can be read-only memory, magnetic disk, or optical disk, etc.

[0141] The above has introduced in detail a computer device provided by the present invention. For those of ordinary skill in the art, according to the idea of the embodiments of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for calculating blood flow pressure, characterized in that: include: S1: Acquire the patient's MRI acceleration encoding image; S2: Inputting the acceleration-coded image into a pressure model to calculate and obtain blood flow pressure; wherein the pressure model is constructed based on the pixel-pressure relationship by calculating the amplitude and pressure gradient in the acceleration-coded image to obtain the pixel-pressure relationship; The method for constructing the pressure model is: S101: Acquire an MRI acceleration encoding image of a patient; S102: Acquire amplitude information based on the acceleration encoded image to obtain an amplitude image; S103: Acquire phase information based on the acceleration encoded image to obtain a phase image; S104: Acquire the pressure gradient of each pixel based on the pixel points of the phase image; S105: Calculating the amplitude image and the pressure gradient to obtain a three-dimensional pressure gradient data set; S106: Obtaining a pixel pressure relationship based on the pressure gradient three-dimensional data set; S107: constructing a pressure model based on the pixel-pressure relationship.

2. The method for calculating blood flow pressure according to claim 1, characterized in that: The pixel pressure relationship is the relationship between the pixel pressure value and the pixel pressure gradient.

3. The method for calculating blood flow pressure according to claim 1, characterized in that: The calculation in step S105 is a dot product calculation.

4. The method for calculating blood flow pressure according to claim 1, characterized in that: The step S104 is specifically as follows: firstly, the acceleration information of each pixel in the phase image is acquired, and then, based on the acceleration information of each pixel, the pressure gradient of each pixel is calculated.

5. The method for calculating blood flow pressure according to claim 1, characterized in that: The S102 further includes amplitude region delineation, delineating a region of interest on the amplitude image to obtain an amplitude region image, and calculating the amplitude region image and the pressure gradient to obtain a three-dimensional data set.

6. The method for calculating blood flow pressure according to claim 5, characterized in that: The amplitude region delineation is to first select the layers containing the region of interest based on the cross-sectional image, perform a first selection on the region of interest of the layers to obtain a selected region image; perform coronal plane reconstruction on the selected image to obtain a coronal plane image, and then perform a second selection on the coronal plane image to obtain an amplitude region image.

7. The method for calculating blood flow pressure according to claim 5, characterized in that: The amplitude region image also includes background removal, which sets the pixels within the region of the amplitude region image to 1 and the pixels outside the region to 0.

8. The method for calculating blood flow pressure according to claim 1, characterized in that: The objective function E(P) of the pixel pressure relationship is expressed as: Among them, φ i,j,k represents the pressure gradient between pixel points (i,j,k) and (i+1,j,k); ψ i,j,k Represents the pressure gradient between pixel points (i,j,k) and (i,j+1,k); Θ i,j,k Represents the pressure gradient between pixel points (i,j,k) and (i,j,k+1), P i,j,k Represents the pressure value of pixel (i, j, k), where M, N, and L represent different dimensions.

9. The method for calculating blood flow pressure according to claim 1, characterized in that: The calculation is to calculate the pressure model by using a conjugate gradient algorithm to obtain the blood flow pressure.

10. The method for calculating blood flow pressure according to claim 9, characterized in that: The conjugate gradient algorithm first derives the objective function of the pressure model to obtain a gradient function, and the gradient function is iteratively calculated until the function converges to obtain the blood flow pressure.

11. The method for calculating blood flow pressure according to claim 10, characterized in that: The iterative calculation is iterated by updating the step size, solution vector, residual and search direction.

12. The method for calculating blood flow pressure according to claim 11, characterized in that: The step size is determined by calculating the objective function and the gradient to update the step size.

13. The method for calculating blood flow pressure according to claim 11, characterized in that: The step size is updated by the following formula, expressed as: α=r T r / p T Ap Among them, α represents the update step size, r represents the residual vector of the current solution P, A represents the coefficient matrix, and p represents the current search direction vector.

14. The method for calculating blood flow pressure according to claim 11, characterized in that: The direction is updated by the following formula, expressed as: p=r+(r new T *r new / r old T *r old )p Among them, p represents the update direction, r represents the residual vector of the current solution P, and r new represents the updated residual vector sum of squares, r old Represents the residual vector sum of squares of the previous iteration, r new T Represents r new The transpose of old T Represents r old The transpose of .

15. The method for calculating blood flow pressure according to claim 1, characterized in that: The method of acquiring the patient's MRI acceleration encoding image includes one or more of the following: 4D FLOW MRI, 3D FLOW MRI, and 2D FLOW MRI.

16. The method for calculating blood flow pressure according to claim 1, characterized in that: The S1 is replaced by acquiring a velocity-encoded image of the patient, and calculating an acceleration-encoded image based on the velocity-encoded image.

17. The method for calculating blood flow pressure according to claim 16, characterized in that: The velocity-encoded image is used to obtain the acceleration-encoded image through differential calculation.

18. The method for calculating blood flow pressure according to claim 1, characterized in that: The pressure calculation of the method also includes any one or more of the following: portal vein pressure, intracranial pressure.

19. A computer program product having a computer program or instructions thereon, characterized in that: The computer program or instructions are executed by a processor to implement the method for calculating blood flow pressure according to any one of claims 1 to 18.

20. A computer device comprising a memory, a processor and a computer program or instruction stored in the memory, characterized in that: include: The processor executes a computer program or instruction to implement the method for calculating blood flow pressure described in any one of claims 1-18.

21. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: The computer program or instructions are executed by a processor to implement the method for calculating blood flow pressure according to any one of claims 1 to 18.

Images