基于血管内超声图像序列的自动三维边缘检测

血管内超声成像可以显示血管内腔、管壁清楚的实时截面图像,从而使定量测量血管参数成为可能。传统手动追踪内腔和管壁边界的方法既费时又存在较大的主观误差,本文将二维主动轮廓算法模型向三维扩展,提出一种适合血管内超声图像的自动三维边缘检测方法,通过对初始给定三维模型实施三维主动界面变形,从而使其逐步逼近目标轮廓。实验结果发现自动检测的内腔、血管壁面积与手动追踪非常吻合,具有较高的相关系数和较小的系统误差,并且纠正二维检测中带来的局部偏差,可作为医生可靠而准确的诊断工具。

Automated Three-Dimensional Contour Detection for Intravascular Ultrasound Image Sequences

Intravascular ultrasound can provide clear real-time cross-sectional images with lumen and plaque.Quantitative measurement can be made by these images.Traditionally,the luminal border and medial-adventitial border are manually traced.This process is extremely time-consuming and the subjective difference is large.In this paper,a new automated three-dimensional contour detection method is introduced for intravascular ultrasound images.It is the fast active contour extended to three dimensions.The initialized three-dimensional model became close to the object by put on active surface deformable model.Experimental results show that luminal and vessel area measurements after automated detection are in agreement with manual tracings with high correlation coefficients and small system difference.Besides,it can correct the local error introduced by fast active contour.And the method provides a reliable and accurately diagnostic tool for doctors.