Abstract:Prediction of the blood flow characteristics is of utmost importance for understanding the behavior of the blood arterial network, especially in the presence of vascular diseases such as stenosis. Computational fluid dynamics (CFD) has provided a powerful and efficient tool to determine these characteristics including the pressure and velocity fields within the network. Despite numerous studies in the field, the extremely high computational cost of CFD has led the researchers to develop new platforms including Machine Learning approaches that instead provide faster analyses at a much lower cost. In this study, we put forth a Deep Neural Network framework to predict flow behavior in a coronary arterial network with different properties in the presence of any abnormality like stenosis. To this end, an artificial neural network (ANN) model is trained using synthetic data so that it can predict the pressure and velocity within the arterial network. The data required to train the neural network were obtained from the CFD analysis of several geometries of arteries with specific features in ABAQUS software. Blood pressure drop caused by stenosis, which is one of the most important factors in the diagnosis of heart diseases, can be predicted using our proposed model knowing the geometrical and flow boundary conditions of any section of the coronary arteries. The efficiency of the model was verified using three real geometries of LAD's vessels. The proposed approach precisely predicts the hemodynamic behavior of the blood flow. The average accuracy of the pressure prediction was 98.7% and the average velocity magnitude accuracy was 93.2%. According to the results of testing the model on three patient-specific geometries, model can be considered as an alternative to finite element methods as well as other hard-to-implement and time-consuming numerical simulations.

Recognition of an obstacle in a flow using artificial neural networks

Flow Reconstruction with Uncertainty Quantification from Noisy Measurements Based on Bayesian Physics-Informed Neural Networks

Hydrodynamic object identification with artificial neural models

A scalable convolutional neural network approach to fluid flow prediction in complex environments

Artificial Neural Networks for Gas‐Liquid Flow Regime Classification in Small Channels

Active flow control with rotating cylinders by an artificial neural network trained by deep reinforcement learning

Development of artificial neural networks for the prediction of the pressure field along a horizontal pipe conveying high-viscosity two-phase flow

Neural Network Model To Predict Slug Frequency of Low-Viscosity Two-Phase Flow

Non-intrusive classification of gas-liquid flow regimes in an S-shaped pipeline riser using a Doppler ultrasonic sensor and deep neural networks

ANN Based Data Integration for Multi-Path Ultrasonic Flowmeter

Classification of flow regimes using a neural network and a non-invasive ultrasonic sensor in an S-shaped pipeline-riser system

Generalization techniques of neural networks for fluid flow estimation

A Flow Feature Detection Method for Modeling Pressure Distribution Around a Cylinder in Non-Uniform Flows by Using a Convolutional Neural Network

Inference of Gas-liquid Flowrate using Neural Networks

Alternative Artificial Neural Network Structures for Turbulent Flow Velocity Field Prediction

Machine-Learning Identification of Hemodynamics in Coronary Arteries in the Presence of Stenosis

An AI-based Domain-Decomposition Non-Intrusive Reduced-Order Model for Extended Domains applied to Multiphase Flow in Pipes

Determination of the Inside Diameter of Pressure Pipes for Drinking Water Systems Using Artificial Neural Networks

Artificial neural network (ANN) modelling to estimate bubble size from macroscopic image and object features

Computable Turbulence Modeling of Laminar-Turbulent Transition Characterized Boundary Layer Flows with the Aid of Artificial Neural Network

Machine Learning Identification Framework of Hemodynamics of Blood Flow in Patient-Specific Coronary Arteries with Abnormality