摘要
医学图像配准是医学图像分析的基本课题之一,具有重要的理论研究和临床诊断应用价值,可以视为参数最优化问题进行求解.作为经典的最优化算法--遗传算法已经在该领域得到了成功应用,然而,遗传算法存在着搜索空间受限、局部搜索能力较差、容易早熟收敛等不足.因此,本文提出了多种策略对遗传算法进行改进,首先结合混沌系统和频度记忆对随机数和随机个体的产生机制进行改进,进而设计一种动态的基因多样性控制器,最后模拟三系杂交水稻的育种机制来发挥杂种优势.同时,结合多分辨率策略实现了多模医学图像的精确配准.公开数据集实验结果表明,相较于散射搜索法、珊瑚礁优化算法、生物地理学优化算法等,本文方法误配率低,鲁棒性强,在相同的时间限制下配准精度高,是一种高效鲁棒的医学图像配准方法.
Being one of the basic topics in medical image analysis,medical image registration has important value in theoretical research and clinical diagnostic application,which is a parameter optimization issue in mathematics. Genetic algorithm,known as a classic optimization algorithm,has been successfully applied in this field. However,there are still disadvantages such as limited search space,poor local search ability,and a high risk of premature convergence. Therefore,this study proposes a variety of strategies to improve the genetic algorithm. Firstly,the chaotic system and frequency-based memory are employed to improve the generation mechanism of random numbers and individuals. Then,a dynamic gene diversity controller is designed. Finally,inspired by heterosis,this study simulates the breeding process of Chinese three-line hybrid rice. Besides,combined with the multi-resolution strategy,genetic algorithm is adapted to multimodal medical image registration successfully. Experiments on public datasets are carried out. Compared with methods such as scatter search method,coral reef optimization algorithm,and biogeography-based optimization,the proposed method has a low mismatch rate,strong robustness,and the most excellent registration accuracy under the same time limit. It is an efficient and robust medical image registration method.
作者
秦泽青
叶志伟
王泽松
徐川
曹羽
QIN Ze-qing;YE Zhi-wei;WANG Ze-song;XU Chuan;CAO Yu(School of Computer Science,Hubei University of Technology,Wuhan 430068,China;School of Computer Science and Information Engineering,Hubei University,Wuhan 430062,China;College of Computer Science and Electronic Engineering,Hunan University,Changsha 410012,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2021年第12期2600-2606,共7页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61772180,61602162)资助
湖北省科技创新项目(2019AAA04)资助
大学生创新创业训练计划项目(201810500001)资助。
关键词
图像配准
遗传算法
混沌
多样性控制
杂种优势
多分辨率
image registration
genetic algorithm
chaos
diversity control
heterosis
multi-resolution
