摘要
提出一种用于CT图像引导脊柱外科系统中,把患者物理坐标系和术前CT图像坐标系进行精确配准的方法.根据临床实际情况,在不依赖激光定位系统的条件下,先从图像和脊椎骨上选择三对对应点配准,得到一个初始稳定解;再通过阈值分割得到脊椎骨的表面点云,在初始解基础上,从脊椎骨表面选取更多点进行迭代最近点优化.应用共线性避免选点带来的初始配准误差,保证配准的可靠性,通过线性距离图提高配准速度.实验结果表明,该方法可在低成本条件下,实现精确便捷配准.
A method was presented for accurate registration between patient coordinate system and CT image coordinate system in CT image guided spine surgery. In a clinical environment without laser localizer, first a corresponding point-pairs landmark registration was used as an initial stable solution from the three point-pairs chosen from the CT image and their corresponding vertebra. Then the threshold method was used to extract the surface point of vertebra. Finally the registration was refined from coarse to fine level with more points chosen from outer surface of vertebra based on iterative closest point principle. A collinear test was used to reduce initial registration error due to initial mischosen points. A distance map was introduced to speed up the registration. An experiment shows that the registration can be implemented accurately, conveniently and quickly in a cost-effective way.
出处
《深圳大学学报(理工版)》
EI
CAS
北大核心
2007年第4期406-409,共4页
Journal of Shenzhen University(Science and Engineering)
基金
粤港重点领域突破资助项目(2005A11304003)
关键词
图像配准
迭代最近点
共线性
距离图
图像引导脊柱外科
image registration
iterative closest point
collinear test
distance map
image guided spine surgery