Computer-Assisted Preoperative Planning for Reduction of Proximal Femoral Fracture Using 3-D-CT Data

Computer-Assisted Preoperative Planning for Reduction of Proximal Femoral Fracture Using 3-D-CT Data
复制标题

DOI:
10.1109/tbme.2008.2005970
复制
发表时间:
2009-03-01
影响因子:
4.6
通讯作者:
Sato, Yoshinobu
Sato, Yoshinobu
中科院分区:
工程技术2区
文献类型:
--
作者:
Okada, Toshiyuki;Iwasaki, Yuta;Sato, Yoshinobu

文献摘要

被引文献

相似文献

本文介绍了使用三维计算机断层扫描(CT)重新定位计算骨折碎片的程序,旨在术前规划计算机引导股骨近端骨折复位。骨碎片的骨折边界,作为“骨折线(FL)”,以及从3-D-CT提取的镜像转换的对侧股骨形状用于重新定位碎片。我们首先描述了一种基于3-D曲率分析提取FL的方法,然后使用以下三个约束条件制定基于骨碎片配准的重新定位方法:1)对侧(Cl.)股骨形状; 2)FL;以及3)CL股骨形状和骨折线,作为“两种约束”。我们使用来自5名模拟和4名真实的股骨近端骨折患者的CT数据集进行实验。我们评估了重新定位计算的旋转误差和重新定位的碎片边界之间的接触率,这分别是恢复正确的功能轴和碎片的骨粘附的关键。实验结果表明,在使用边界线约束进行配准后,再使用两种约束进行配准,具有良好的精度和稳定性。平均而言,6.0度+/- 0.8度的旋转误差和89% +/- 3%的接触率,而无需提供精确的初始值。
This paper describes procedures for repositioning calculations of fractured bone fragments using 3-D-computed tomography (CT), aimed at preoperative planning for computer-guided fracture reduction of the proximal femur. Fracture boundaries of the bone fragments, as "fracture lines (FLs)," and the mirror-transformed contralateral femur shape extracted from 3-D-CT were used for repositioning of the fragments. We first describe a method for extracting FLs based on 3-D curvature analysis and then formulate repositioning methods based on registration of bone fragments using the following three constraints: 1) contralateral (Cl.) femur shape; 2) FLs; and 3) both CL femur shape and fracture lines, as "both constraints". We performed experiments using CT datasets from five simulated and four real patients with proximal femoral fracture. We evaluated the rotation error in reposition calculations and the contact ratio between repositioned fragment boundaries, which are crucial for the recovery of proper functional axes and bone adhesion of fragments, respectively. Experimental results showed that good accuracy and stability were attainable when registration using both constraints was performed after registration using the fracture-line constraint. On average, 6.0 degrees +/- 0.8 degrees in rotation error and 89% +/- 3% in contact ratio were obtained without providing precise initial values.