A Physics-Informed Neural Network Approach to the Pulse Wave Inverse Problem for Noninvasive Intravascular Pressure Estimate

The aim of the study is to utilize vessel wall displacements and blood flow velocities obtained with Pulse Wave Imaging (PWI) and Vector Flow Imaging (VFI) techniques to estimate wall compliance and intraluminal pressure using the Pulse Wave Inverse Problem (PWIP) approach. The PWIP method was employed combining numerical approach and physics-informed neural network (PINN) to optimize parameter estimation. An ultrasound simulation framework combining fluid-structure interaction (FSI) simulations with FIELD-II simulations provided ultrasound data and pressure truth were used for validation and optimization of PWIP. Results demonstrated strong correlation and minimal bias in intraluminal pressure and wall compliance estimations using PWIP.