J0210104 複数医用画像を用いた動脈壁モデルの構築

Abstract

A method for coronary arterial dynamics computation with medical-image-based time-dependent anatomical models was introduced. One of the important components of that method is a curve-extraction technique from time-dependent medical images. The curve-extraction technique is based on minimizing the strain energy for a curved-beam representation of the artery, and from that the arterial model is constructed by associating time-averaged cross-sections to the points along the curve. Here we present a new method for arterial wall modeling from time-dependent medical images. We have developed a surface-extraction technique for obtaining the time-dependent anatomical model from the medical images. In this technique, the arterial surface geometry is extracted from the medical image by using a NURBS representation of the extracted surface, and the surfaces associated with different instants of the cardiac cycle all have a common parametric space. In this technique, a reference is constructed using NURBS. The NURBS representation of the surface becomes a target geometry in the strain-energy minimization, and the resulting equations are solved over the arterial surface. We apply the method to a human aorta, with 20 images per cardiac cycle.