Neural Network Blood Flow Field Prediction from Vascular Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods lack an efficient and accurate way to determine blood flow field information for vascular segments, which is crucial for assessing vascular health and guiding medical interventions.
Innovation Solution
A method using a pre-trained neural network to analyze vascular segment images, predicting physical quantities like pressure, velocity, and flow rate that satisfy physical constraints, and determining human medical information such as FFR or plaque risk based on these predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods are used to determine blood flow field information, then the process is simpler, but the accuracy and comprehensiveness of the assessment is insufficient
Solution Approach 1:
The patent replaces traditional mechanical or manual blood flow assessment methods with a neural network-based computational system. The neural network automatically processes vascular segment images to predict multiple physical quantities (pressure, velocity, flow rate) that satisfy physical constraints, thereby improving measurement precision while managing complexity through automated computation.
Solution Approach 2:
The patent transforms the blood flow field information determination by predicting multiple physical quantity parameters (pressure, velocity, flow rate) simultaneously through the neural network. This multi-parameter prediction approach enhances the comprehensiveness and accuracy of the assessment compared to traditional single-parameter methods.
2Reliability
If multiple physical quantities are predicted to improve diagnostic accuracy, then the understanding of blood flow dynamics is enhanced, but the computational complexity increases
Solution Approach 1:
The neural network is designed to perform multiple functions simultaneously: predicting pressure distribution, velocity fields, and flow rates within a single unified model. This multi-functional approach enhances diagnostic reliability by providing comprehensive blood flow field information while managing computational complexity through integrated processing.
Solution Approach 2:
The patent incorporates physical constraint conditions as feedback mechanisms to guide the neural network's predictions. The predicted physical quantities are constrained to satisfy fundamental physical laws (e.g., conservation of mass, Bernoulli's principle), ensuring that the enhanced diagnostic accuracy is achieved while maintaining physical realism and managing computational complexity.
3Reliability
If physical constraint conditions are imposed on predicted values, then the physical realism is improved, but the flexibility of the prediction model is reduced
Solution Approach 1:
Physical constraint conditions are implemented as feedback mechanisms that guide the neural network's prediction process. The model learns to satisfy these constraints during training, incorporating physical laws (such as conservation of mass and energy) into the prediction framework. This approach maintains physical realism while the neural network's learning capability preserves model flexibility and adaptability to different vascular configurations.
Data Source
Figure 1
Figure 2~3A
Figure 3B~3C
AI summary
A method, device, computing equipment, and storage medium for determining blood flow field information are provided. The method can include: obtaining vascular segment images regarding a target vascular segment; and based on the vascular segment images, determining at least one blood flow field information for the said target vascular segment through a pre-trained neural network.