基于ITK的多模态医学图像非刚性配准研究

Non-rigid Registration of Multimodality Medical Images Based on ITK

目的:研究基于the insight toolkit(简称ITK)的多模态医学图像非刚性配准,分析基于ITK的多模态医学图像非刚性配准效果。方法:从医学图像配准的基本框架出发,针对多模态医学图像的非刚性配准,深入研究了框架各模块在配准中应发挥的作用,选择了利用互信息作为相似性测度、Blackman-Harris窗函数改进的部分体积插值(PV)、基于B样条的自由形变变换方式、有限存储LBFGS搜索算法,以满足多模态非刚性图像配准要求。结果:利用ITK软件包,实现了对人体肝脏的核磁和CT图像的配准。结论:基于改进的医学图像配准框架实验数据证明了该方法的有效性。

Objective To realize and analyze the registration of multimodality and non-rigid images by Insight Toolkit（ITK）.Methods Based on basic frame of medical image registration and non-rigid registration of multi-mode medical image,the functions of all modules in the registration were studied.Also,it＇s introduced how the modules worked in the registration frame.The normalized mutual information model was integrated with improved Partial Volume Interpolation model,and Blackman-Harris Window kernel function was used for multimodality image registration.The model of Free-Form Deformation（FDD） based on B-spline and the optimizer named Limited memory（LBFGS） were used to accommodate non-rigid image transformation.Results The registration of liver MR and CT images was tested by the ITK.Conclusion Experimental data shows the validity of the proposed metho.