Elastic Registration for Medical Image Based on Progressively Fine Feature Extraction
Yu Shi · Journal of Lanzhou Jiaotong University · 2015
To overcome the shortcoming of registration based on feature point for large scale deformation 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 preliminary 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.