Doppler-based 3D Blood Flow Imaging and Visualization

Abstract

Blood flow is a very important part of human physiology. In this paper, we present a new method for estimating and visualizing 3D blood flow on-the-fly based on Doppler ultrasound. We add semantic information about the geometry of the blood vessels in order to recreate the actual velocities of the blood. Assuming a laminar flow, the flow direction is related to the general direction of the vessel. Based on the center line of the vessel, we create a vector field representing the direction of the vessel at any given point. The actual flow velocity is then estimated from the Doppler ultrasound signal by back-projecting the velocity in the measured direction, onto the vessel direction. Additionally, we estimate the flux at user-selected cross-sections of the vessel by integrating the velocities over the area of the cross-section. In order to visualize the flow and the flux, we propose a visualization design based on traced particles colored by the flux. The velocities are visualized by animating particles in the flow field. Further, we propose a novel particle velocity legend as a means for the user to estimate the numerical value of the current velocity. Finally, we perform an evaluation of the technique where the accuracy of the velocity estimation is measured using a 4D MRI dataset as a basis for the ground truth.