CN-122025152-A - Fluid parameter prediction method, device, apparatus, medium, and program product

CN122025152ACN 122025152 ACN122025152 ACN 122025152ACN-122025152-A

Abstract

The disclosure provides a fluid parameter prediction method, a device, equipment, a medium and a program product, and relates to the technical field of data processing. The method comprises the steps of obtaining medical image data and physiological information data of an object to be predicted, wherein the medical image data are used for representing structural characteristics of a target part of the object to be predicted, the physiological information data are used for representing physiological state characteristics of the object to be predicted, inputting the medical image data and the physiological information data into a trained fluid parameter prediction model, outputting fluid parameters of the target part at different moments in a full cardiac cycle, and the fluid parameter prediction model is constructed based on a deep learning model and is obtained through training by a data driving loss function with physical constraints. The method solves the problem of prediction deviation when the fluid parameter is predicted by singly using pure image data or pure physiological parameters, and realizes accurate and rapid prediction of the fluid parameter at different moments in the full cardiac cycle of the target part.

Inventors

WANG JINGQI
TANG ZICHUN
HU JIA
CHEN HONGGANG
MA JIAXING

Assignees

四川大学华西医院
防城港梅桥医疗科技有限公司

Dates

Publication Date: 20260512
Application Date: 20251011

Claims (11)

1. A method of predicting a fluid parameter, comprising: Acquiring medical image data and physiological information data of an object to be predicted, wherein the medical image data is used for representing structural characteristics of a target part of the object to be predicted, and the physiological information data is used for representing physiological state characteristics of the object to be predicted; Inputting the medical image data and the physiological information data into a trained fluid parameter prediction model, and outputting fluid parameters of the target part at different moments in the whole cardiac cycle, wherein the fluid parameter prediction model is constructed based on a deep learning model and is obtained by training a data driving loss function with physical constraints.
2. The method of claim 1, wherein the medical image data comprises aortic geometry data acquired based on computed tomography angiography, CTA, and the physiological information data comprises at least one of blood pressure, heart rate, cardiac output, and systolic duty cycle.
3. The method of predicting a fluid parameter as set forth in claim 1, wherein the target site is an aorta and the fluid parameter comprises at least one of wall shear stress WSS, time-averaged wall shear stress TAWSS, oscillation shear index OSI, relative residence time RRT, and helicity Psi.
4. The fluid parameter prediction method according to claim 1, wherein acquiring medical image data and physiological information data of the object to be predicted includes: Performing three-dimensional modeling and grid division on the medical image data to generate a three-dimensional model containing the geometric boundary of the target part; And inputting the physiological information data as boundary conditions into Computational Fluid Dynamics (CFD) simulation, and generating fluid parameter reference data corresponding to the medical image data.
5. The method for predicting a fluid parameter as set forth in claim 1, wherein the training process of the fluid parameter prediction model comprises: Converting the acquired medical image data into 3D point cloud data, and constructing a training data set by combining the physiological information data; Inputting the training dataset into a deep learning-based model framework comprising Kpconv encoder-decoder structures capable of learning multi-scale features, and a graph convolutional neural network for mining point cloud context information; configuring a data driving loss function with physical constraint on the model framework, wherein the physical constraint is constructed based on a continuity equation and a Navier-Stokes equation, and the data driving loss function is used for measuring errors between a model prediction result and reference data obtained by computational fluid dynamics simulation; And training the model framework by adopting an adaptive moment estimation optimizer Adam to obtain the fluid parameter prediction model.
6. The fluid parameter prediction method of claim 5, further comprising one or more of the following prior to inputting the training dataset into a deep learning based model framework: Processing the 3D point cloud data by adopting a furthest point sampling FPS algorithm; applying a random rigid transformation to the 3D point cloud data, including random rotation and random translation; randomizing the arrangement sequence of the points in the 3D point cloud data.
7. The method for predicting a fluid parameter as set forth in claim 1, wherein outputting the fluid parameter at different times in the full cardiac cycle of the target site comprises: predicting flow field parameters corresponding to point by point through the fluid parameter prediction model, wherein the flow field parameters comprise speed components and pressure scalar quantities in three directions; and calculating the distribution of the hemodynamic parameters of the target part at different moments based on the flow field parameters.
8. A fluid parameter prediction apparatus, comprising: The data acquisition module is used for acquiring medical image data and physiological information data of an object to be predicted, wherein the medical image data is used for representing structural characteristics of a target part of the object to be predicted, and the physiological information data is used for representing physiological state characteristics of the object to be predicted; And the parameter output module is used for inputting the medical image data and the physiological information data into a trained fluid parameter prediction model and outputting fluid parameters of different moments of the whole cardiac cycle of the target part, and the fluid parameter prediction model is constructed based on a deep learning model and is obtained by training a data driving loss function with physical constraint.
9. An electronic device, comprising: Processor, and A memory for storing executable instructions of the processor; Wherein the processor is configured to perform the fluid parameter prediction method of any one of claims 1-7 via execution of the executable instructions.
10. A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the fluid parameter prediction method of any one of claims 1 to 7.
11. A computer program product comprising a computer program or instructions which, when executed by a processor, implements the fluid parameter prediction method of any one of claims 1 to 7.

Description

Fluid parameter prediction method, device, apparatus, medium, and program product Technical Field The present disclosure relates to the field of data processing technologies, and in particular, to a fluid parameter prediction method, a device, an apparatus, a medium, and a program product. Background In the fields of biomedical engineering and clinical medical treatment, deep research on internal fluid dynamics of human body is important for disease diagnosis, treatment scheme formulation and medical instrument research and development. The fluid parameters are used as key indexes reflecting the physiological states and hydrodynamic characteristics of human bodies, and the acquisition and analysis of the fluid parameters have irreplaceable roles in understanding physiological processes such as cardiovascular systems and the like. For example, in the research of cardiovascular diseases, the fluid parameters in the aorta can provide important basis for doctors to judge the risk of vascular diseases and evaluate the heart functions, and in the research and development of medical instruments, the accurate fluid parameters are helpful for optimizing the design of ventricular assist devices, artificial heart valves and other products and improving the performance and safety of the ventricular assist devices, artificial heart valves and other products. However, there are currently significant drawbacks in the technique of acquiring fluid parameters. The existing measurement and prediction methods are difficult to meet the clinical and scientific research requirements, and mainly face double challenges of efficiency and accuracy. On the one hand, the traditional measurement technology is often complicated in operation and long in time consumption, cannot realize rapid detection, and cannot provide critical fluid parameter information for doctors in time when facing emergency medical conditions, so that the best opportunity for diagnosis and treatment of illness states can be delayed. On the other hand, the existing prediction model is uneven in accuracy and is interfered by various factors, and the prediction result has larger deviation from the actual situation, so that the reliability of the prediction model in clinical application and scientific research analysis is greatly reduced. Therefore, how to quickly and accurately obtain the fluid parameters becomes a key technical problem to be solved. It should be noted that the information disclosed in the above background section is only for enhancing understanding of the background of the present disclosure and thus may include information that does not constitute prior art known to those of ordinary skill in the art. Disclosure of Invention The present disclosure provides a fluid parameter prediction method, apparatus, device, medium, and program product that overcome, at least to some extent, the problems of low fluid parameter prediction efficiency and low accuracy. Other features and advantages of the present disclosure will be apparent from the following detailed description, or may be learned in part by the practice of the disclosure. According to one aspect of the disclosure, a fluid parameter prediction method is provided, and the fluid parameter prediction method comprises the steps of obtaining medical image data and physiological information data of an object to be predicted, wherein the medical image data are used for representing structural characteristics of a target part of the object to be predicted, the physiological information data are used for representing physiological state characteristics of the object to be predicted, inputting the medical image data and the physiological information data into a trained fluid parameter prediction model, outputting fluid parameters of the target part at different moments of a full cardiac cycle, and the fluid parameter prediction model is constructed based on a deep learning model and is obtained through training through a data driving loss function with physical constraints. In some embodiments, the medical image data comprises aortic geometry data acquired based on computed tomography angiography CTA, and the physiological information data comprises at least one of blood pressure, heart rate, cardiac output, and systolic duty cycle. In some embodiments, the target site is the aorta and the fluid parameter includes at least one of wall shear stress WSS, time-averaged wall shear stress TAWSS, oscillation shear index OSI, relative residence time RRT, and helicity Psi. In some embodiments, acquiring medical image data and physiological information data of an object to be predicted comprises three-dimensionally modeling and meshing the medical image data to generate a three-dimensional model containing the geometric boundary of the target part, and inputting the physiological information data as boundary conditions into Computational Fluid Dynamics (CFD) simulation to generate fluid parameter reference data corr