一种改进的条件方差和医学图像配准  

An improved medical image registration method based on the sum of conditional variance

作　　者：相艳[1] 桂鹏 王硕 许春荣 邵党国[1] 刘利军[1] 汤守国[1] XIANG Yan;GUI Peng;WANG Shuo;XU Chunrong;SHAO Dangguo;LIU Lijun;TANG Shouguo(College of Electronic Information and Automation,Kunming University of Science and Technology,Kunming 650051,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,昆明650051

出　　处：《生物医学工程研究》2018年第1期71-76,共6页Journal Of Biomedical Engineering Research

基　　金：国家自然科学基金资助项目(81560296);云南省教育厅科学研究基金资助项目(2015Y070)

摘　　要：多模医学图像配准是将不同医学成像模式提供的影像信息进行融合的关键步骤。条件方差和(SCV)是一种新的用于多模图像配准的相似性测度,但SCV的主要缺点是它仅使用量化信息来计算联合直方图。基于此,设计了一种新的插值函数来计算联合直方图,从而提高SCV的性能。将改进后的SVC用于多模医学图像配准,并与归一化互信息(MI)、交叉累积剩余熵(CCRE)和原SCV进行了比较。实验证明,相比NMI、CCRE和原SCV,本研究方法能配准具有不同空间变换和噪声的图像,具有更高的配准成功率和鲁棒性。Multi-modality medical image registration is the key step for the fusion of image information provided by different medical imaging modalities.The sum of conditional variance(SCV)is considered to be a state-of-the-art similarity measure for registering multi-modality images.However,the main drawback of the SCV is that it uses only quantized information to calculate the joint histogram.To overcome this limitation,we proposed a new interpolator function for the joint histogram of SCV to improve the performance.We applied our method to multi-modality medical image registration,comparing with the normalized mutual information(NMI),cross cumulative residual entropy(CCRE),and the original SCV.The results show that the proposed method can register images with different geometric transformation or strong noises.Compared with NMI,CCRE and SCV,the proposed method has improved the success rate and robustness.

关 键 词：图像配准 条件方差和 PV插值 归一化互信息 交叉累积剩余熵 

分 类 号：R318[医药卫生—生物医学工程] TP391[医药卫生—基础医学]