基于多分辨率策略的弹性配准方法

Elastic Registration Method Based on Multi-Resolution Strategy

提出了一种基于多分辨率策略的医学图像弹性配准方法。首先通过使用多分辨率C-V模型来加快边界提取的速度,接着采用模拟退火结合薄板样条插值的方法来配准边界点。为了增强边界点配准算法的鲁棒性和速度,在退火过程中使用多分辨率策略来构造点集数据。最后利用薄板样条插值函数完成弹性变换。实验结果表明,算法对于大形变的情况依然具有很好的效果。

A hybrid elastic registration method based on multi-resolution strategy for medical images is presented. Firstly, the C-V model is used to extract contours from images, and its speed is accelerated by using multi-resolution strategy. Then two contour point sets are matched by an annealing procedure with thin-plate spline interpolation. To enhance the robustness of the matching algorithm and accelerate the speed, another multi-resolution strategy is used to construct contour point sets. Finally, the elastic transformation using smooth thinplate spline interpolation is performed. Experimental results demonstrate that the algorithm has good performance even in some large deformation cases.