基于结构光扫描的椎弓根多区域组合配准技术

Multi-segmentation image associated registration technology for pedicle screw insertion based on structured light scanning

目的利用术中结构光扫描图像和术前CT进行配准,实现计算机辅助椎弓根手术。方法术前对手术部位的CT图像进行三维重建及区域划分,术中用结构光对其进行三维扫描,把腰椎模型骨各区域组成17个组合,然后结合预配准与多区域ICP算法对两点集进行配准。结果该算法实现了脊椎骨可见光数据与CT数据的配准,即使在噪声干扰和图像偏移的影响下,配准误差仍能保证在1mm之内。结论本文提出的预配准结合多区域配准算法能有效地避免ICP算法的局部收敛性;即使在施加噪声和外部干扰情况下,利用两个区域组合进行配准可以满足手术精度要求。

Objective To register intraoperative structured light images with preoperative CT images in computer aided orthopaedic surgery. Methods CT images were obtained preoperatively and divided into different patches, then these point sets were combined into 17 combinations. Pre-registration and multi-segmentation ICP algorithm were performed to register them to vertebral surface points acquired intraoperatively by structured light scanning. Results Even with noise and per- turbation to the structured light images of the spine plastic phantom, registration error was less than 1 ram. Conclusion A new registration method for spine surgery based on structured light scanning with multi-segmentation image is presented in this paper, which prevents ICP algorithm from converging to a local minimum, and increases the accuracy in pedicle screw insertion surgery. Submillimetric registration accuracy can be obtained even when a small portion of the posterior vertebral surface is used for registration.