基于互严格凹函数测度的医学图像配准新方法  

Medical Image Registration Based on Mutual Strictly Concave Function Measures

作　　者：刘常春[1] 胡顺波[1] 俞梦孙[2] 

机构地区：[1]山东大学控制科学与工程学院,济南250061 [2]空军航空医学研究所,北京100036

出　　处：《中国生物医学工程学报》2007年第2期231-237,共7页Chinese Journal of Biomedical Engineering

基　　金：国家"863"计划资助项目(2006AA02Z4D9);山东省自然科学基金资助项目(Z2006C05)

摘　　要：为了提高医学图像配准的运算速度和增大配准的稳定区域,提出用严格凹函数取代互信息中的香农熵函数的方法,形成了互严格凹函数测度(NMi,i=1,2,…,6)。互信息只是互严格凹函数测度的一个特例。并提出一种新的判断配准好坏的标准——稳定区域和稳定区域宽度。通过试验比较得NM2、NM4、NM5互严格凹函数测度比互信息好,比I0.2和R0.2测度更好,即这三个测度的配准时间比互信息少,比I0.2和R0.2测度更少;三个测度配准的平均稳定区域宽度比互信息宽,比I0.2和R0.2测度更宽。最后用NM2测度进行多模态医学图像的非刚体配准试验,结果表明效果良好。In order to enhance the medical image registration speed and enlarge the stable region, a number of new registration measures were proposed. The new measures were derived by replacing the Shannon entropy function in mutual information with any strictly concave function, which were named mutual strictly concave function measures （ NMi , i = 1, 2,…, 6）. The mutual information measure is a special case. At the same time this paper gave a new standard of distinguishing the registration results, i.e. the stable region and the stable region width. The test results showed that NM2 , NM4 , NM5 measures were better than the mutual information measure which was better than Ⅰ0.2 and R0.2 measures. The three new measures expended less time than the mutual information, Ⅰ0.2 and R0.2 measures, and they had larger stable region width than other methods. A muhi-modality medical image nonrigid registration test was conducted by NM2 measure and the good result was obtained.

关 键 词：互信息 严格凹函数 医学图像 配准 

分 类 号：R318[医药卫生—生物医学工程]