Exploring Ultrasound Images of the Carotid Arteries Using Neural Network Tools

Abstract

This work aims at 3D model reconstruction and hemodynamic analysis of carotid bifurcations based on hospital ultrasound images. The suggested approach addresses B mode image segmentation as a pattern recognition problem and boundary velocity conditions are extracted from Doppler PW mode velocity data. Both tasks are solved using a combination of artificial neural network (ANN) tools. The development of an integrated technique for geometric reconstruction, structured meshing, blood flow simulation and hemodynamic analysis of the carotid bifurcation enables the quantification of the degree of stenosis and promotes the understanding of how hemodynamic factors are involved in atherosclerotic disease.