Hemodynamic Prediction Using Dual Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computational fluid dynamics methods for diagnosing and monitoring artery diseases such as aneurysms and coronary diseases are costly, complex, and slow for clinical purposes.
Innovation Solution
A machine learning-based system using two artificial neural networks (ANNs) to automatically determine hemodynamic characteristics of blood vessels from 3D anatomical models derived from medical images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If computational fluid dynamics methods are used to compute blood flow characteristics, then diagnostic accuracy is improved, but computational cost and complexity increase significantly
Solution Approach 1:
The patent replaces the mechanical computational fluid dynamics system with a machine learning-based system. Instead of solving complex partial differential equations through numerical methods, the invention uses trained neural networks to predict hemodynamic characteristics directly from anatomical images, substituting the computational mechanics approach with a data-driven machine learning approach that maintains diagnostic accuracy while reducing computational burden
Solution Approach 2:
The patent applies preliminary action by pre-training the machine learning models on extensive datasets of anatomical images and corresponding hemodynamic characteristics. The neural networks are trained in advance using computational fluid dynamics results as ground truth, so that during clinical use, the pre-trained models can rapidly predict hemodynamic characteristics without performing real-time complex computations
2Measurement precision
If computational fluid dynamics methods are used to compute blood flow characteristics, then diagnostic accuracy is improved, but computation time increases
Solution Approach 1:
The patent replaces the time-consuming computational fluid dynamics mechanical computation system with a machine learning prediction system. The neural networks, once trained, can generate hemodynamic characteristic predictions in seconds or minutes rather than hours or days, dramatically reducing computation time while preserving diagnostic accuracy through the learned mappings from anatomical features to hemodynamic parameters
Solution Approach 2:
The patent performs the computationally intensive training phase in advance using pre-collected datasets. During actual clinical application, the pre-trained models execute rapid predictions without requiring real-time complex computations, thus eliminating the time loss associated with on-demand computational fluid dynamics calculations
3Measurement precision
If machine learning models are trained using extensive datasets, then prediction accuracy is improved, but training time and computational resources increase
Solution Approach 1:
The patent applies preliminary action by collecting and preparing extensive training datasets in advance, including anatomical images and corresponding hemodynamic characteristics from computational fluid dynamics simulations. The data preprocessing, augmentation, and model training are performed beforehand, creating ready-to-deploy models that can be rapidly applied to new patient cases without requiring additional training time
Solution Approach 2:
The patent implements dynamic training strategies where the model can be incrementally updated or fine-tuned with new data as it becomes available, rather than requiring complete retraining. This allows the system to adapt to new anatomical variations and improve accuracy over time while minimizing the computational burden of continuous full-model retraining
Data Source
AI summary
Described herein are systems, methods, and instrumentalities associated with automatic determination of hemodynamic characteristics. An apparatus as described may implement a first artificial neural network (ANN) and a second ANN. The first ANN may model a mapping from a set of 3D points associated with one or more blood vessels to a set of hemodynamic characteristics of the one or more blood vessels, while the second ANN may generate, based on a geometric relationship of the set of points in a 3D space, parameters for controlling the mapping. The apparatus may obtain a 3D anatomical model representing at least one blood vessel of a patient based on one or more medical images of the patient, and determine, based on the first ANN and the second ANN, a hemodynamic characteristic of the at least one blood vessel of the patient at a target location of the 3D anatomical model.