基于逐步细化特征提取的医学图像弹性配准

Elastic Registration for Medical Image Based on Progressively Fine Feature Extraction

针对基于特征点的弹性配准方法对局部存在较大尺度形变的医学图像配准精度较低的不足,提出一种从全局到局部逐步细化的特征提取弹性配准方法.该方法首先采用尺度不变特征变换算法(scale invariant feature transform,SIFT)对图像进行特征提取与匹配以完成初步配准.然后通过计算参考图像和初步配准图像中所有控制点的领域均方差(mean square difference,MSD)寻找存在较大尺度形变的局部区域,在这些区域使用互信息再次进行特征提取,并利用层次B样条插值实现图像的精确配准.实验结果表明:与基于互信息的特征提取配准方法相比,该方法在保证配准精度的同时,有效地提高了配准速度.

To overcome the shortcoming of registration based on feature point for large scale de-formation medical image,an elastic registration method has been proposed in this paper.It based on progressively fine feature extraction from global to local.First,the scale invariant feature transform（SIFT）algorithm is used for feature extraction and matching to complete preliminary registration.Then the local region which has a large scale deformation is obtained by calculating mean square difference（MSD）of the field of all control points for the reference image and the pre-liminary registration image.Feature extraction based on mutual information is utilized in those regions,and accurate registration is achieved by hierarchical B-spline interpolation.Experimental results demonstrate that compared with registration method based on feature extraction by mutual information,the speed of the proposed method is faster while registration accuracy is ensured.