摘要
多模图像配准一直是医学图像处理领域研究的热点,而智能优化算法是影响配准算法性能的一个关键因素,很少有学者对常用智能优化算法的特性进行深入的分析。文章以二维CT和PET图像为数据源,互信息为测度函数,分别对禁忌搜索、模拟退火、遗传、进化规划、粒子群以及蚁群等六种智能优化算法进行了实验,具体从配准误差、算法耗时和收敛速度对优化算法的性能进行定量评估,并综合评价它们的优缺点。实验结果表明:文章研究对多模医学图像配准中的优化算法的选择具有较好的参考价值。
Multi-modal image registration has been a hot research field in medical image processing. Intelligent optimi-zation algorithm is one of the key factors affecting the performance of registration. However,there are few articles studying these optimization algorithms by deep comparisons. In this paper,two-dimensional CT and PET images with mutual infor-mation as the similarity metric are used to test the tabu search, simulated annealing, genetic,evolutionary programming,particle swarm and ant colony optimization algorithms respectively. Furthermore,registration error,run time,and conver-gence speed are selected to evaluate these algorithms comprehensively. The experimental results indicate that our study offers a preferable reference in selection of optimization algorithms for multimodal medical image registration.
出处
《计算机与数字工程》
2016年第12期2467-2473,共7页
Computer & Digital Engineering
基金
国家自然科学基金项目(编号:61201117
61301042)
中国科学院青年创新促进会(编号:2014281)
江苏省自然基金项目(编号:BK20151232)
苏州市科技计划项目(编号:ZXY2013001)资助
关键词
多模图像配准
智能优化算法
互信息
优化比较
multi-modal image registration , intelligent optimization algorithm , mutual information , optimization com-pare