一种基于统计参数映射融合独立成分分析的新的.pdf

25 3 JOURNAL OFNATIONAL UNIVERSITYOF DEFENSE TECHNOLOGY Vol.25 No.3 2003 :1001- 2486(2003) 03- 0096- 04 * 1 2 2 3 雷 震 , 胡德文 , 周宗谭 , 朱文珍 (11 , 410073; 21 , 410073; 31 , 430032) : fMRI , , SPM ICA , ICA GLM , : ; ; ; :Q632 :A A New Activation Detecting Method Based on Combination of Statistical Parametric Mapping with the Independent Component Analysis 1 2 2 3 LEI Zhe ,HU De-we ,ZHOU Zong-ta , ZHU Wen-zhe (1. College of Humanities and Management, NationalUniv. of DefenseTechnology, Changsha 410073, China; 2. College ofMechatronics Engineering and Automation, National Univ. ofDefense Technology, Changsha 410073, China; 3. Department ofRadiology, Tong JiHospital, Tong JiMedicalSchool, Huazhong Univ. ofSci. Tech. , Wuha n430032, China) Abstract:Statistical parametricmapping (SPM) depends o nthe general linear model( GLM) and the theory of Gaussia nfields to some extent. But the disadvantages of theGLM are related to the fact that these assumptions outlined do not fairly represent the fMRI data. Also, hemodynamics and distributed patterns of the brai nactivity may not be well modeled by the GLM regressio nframework. While, the independent component analysis( ICA) does not provide the investigator with a significance estimate for each component activation, which may discourage experimenters from attempting to interpret the results. The paper proposes a method which combines some of thebenefits of ICA with thehypothesis-testing approach of theSPM. Experimenta