ICP优化下的B样条四维CT图像弹性配准方法

ICP Optimized B-Spline Non-Rigid Registration for 4D-CT Images

为提高不同生理状态下两组四维CT图像之间配准的精度和速度,基于多分辨率B样条的自由形变模型(Free Form Deformation, FFD),提出一种使用迭代最近点(Iterative Closest Point,ICP)优化该模型的配准算法。在传统B样条之前加入ICP算法实现两组四维CT图像间的点云配准:根据分割完的两组四维CT图像生成点云数据和灰度数据,使用ICP对模型中的两组点云配准。level1,level 2,level 3相似性测度提高率分别为:8.68%,10.46%,2.39%,速度提高率分别为:–51.89%,41.71%,81.09%,结果证明新模型在不同控制网格大小配准上精度和速度都有提高。

In order to improve the accuracy and speed of registration between two groups of 4D-CT images under different physiological activities,based on FFD model of multi-resolution B-spline,a registration algorithm using ICP to optimize the model is proposed.The ICP algorithm is added before the traditional b-spline to realize the point cloud registration between two groups of 4D-CT images:the appropriate model is generated from the two segmented groups of 4D-CT images,then the two groups of point clouds in the model are registered with ICP.The improvement rates of similarity measurement at level 1,level 2 and level 3 are 8.68%,10.46%and 2.39%respectively,and the speed improvement rates are-51.89%,41.71%,81.09%respectively.