Medical image registration based on Clifford algebra and three-dimensional SURF
作者单位:School of AutomationNanjing University of Science and Technology School of Electrical EngineeringNantong University
会议名称:《第32届中国过程控制会议(CPCC2021)》
会议日期:2021年
学科分类:0831[工学-生物医学工程(可授工学、理学、医学学位)] 08[工学] 080203[工学-机械设计及理论] 0802[工学-机械工程] 0836[工学-生物工程]
关 键 词:medical image registration geometric algebra SURF algorithm feature descriptor fast space optimization
摘 要:Multimodal medical image registration can provide doctors with more abundant focus information of patients,but there is a contradiction between registration accuracy and efficiency in clinical *** this paper,a 3 D SURF framework based on Hessian four-dimensional scale space is built,a feature descriptor based on geometric algebraic relative angle invariants is designed,which is more suitable for 3 D images,and a fast spatial optimization(GA-FSO) algorithm based on this descriptor is *** method achieves high precision registration and compensates the lack of immediacy in clinical ***,the registration experiment is carried out based on the brain multimodal images provided by RIRE open database,and the mean error is less than 3 mm,which verifies the effectiveness of this registration scheme and satisfies demand of clinical online *** the contrastive experiments for registration are carried out with clinical case data to verify the clinical practicability of the registration *** study provides theoretical basis and technical support for the clinical application of three-dimensional multimodal medical image registration.