基于小波包技术的脑血流导纳信号熵特征分析  

Feature extraction and analysis using wavelet packets for cerebral admittance plethysomgraphy

作　　者：王磊[1] 童隆正[1] 王卫兵[1] 李训栋[1] 王默力[2] 周萍[1] 

机构地区：[1]首都医科大学生物医学工程学院计算机应用系,北京市100069 [2]首都医科大学宣武医院神经内科,北京市100053

出　　处：《中国临床康复》2006年第13期113-115,共3页Chinese Journal of Clinical Rehabilitation

摘　　要：目的:探讨通过脑血流导纳信号的熵特征分析,揭示脑血管功能状态。方法:选择2004-09/2005-01首都医科大学宣武医院神经内科收治的高血压患者、动脉硬化患者各30例。选择2002-09/2003-01首都医科大学正常者30人作对照。①利用ENG系列导纳式双侧脑血流图自动检测仪,对各组人员进行脑导纳信号的额-乳突导联检测。②利用小波包分解技术提取脑导纳信号不同频段的熵特征,进行对照比较和统计学分析。结果:两组患者各30例,正常对照30人,全部进入结果分析。①小波包2级分解显示,高血压组和动脉硬化组的熵值均低于正常组,高血压组和正常组的T(2,1)熵特征值差异有显著性意义(303.61±20.48,368.09±14.20,P=0.012)。②3级分解显示,在中频段,高血压组脑导纳信号的熵特征值与正常组T(3,1),T(3,2)和T(3,4)差异有显著性意义(P<0.05);而动脉硬化组的熵值差异则全部有显著性意义。结论:利用小波包技术从脑导纳信号中提取的熵特征值正常组和异常组之间存在差异,而且在中频段的特征值存在显著性差异,借助小波包分析手段,通过对脑导纳信号的分解,研究脑血管功能状态,这将为进一步的研究如信号分类等提供依据。AIM： To probe into the feature of cerebral admittance plethysomgraphy, and study cerebrovascular functional state. METHODS：We recruited 30 patients with hypertension and arteriosclerosis respectively who received treatment in Department of Neurology, Xuanwu Hospital, Capital University of Medical Sciences from September 2004 to January 2005. Another 30 healthy persons were chosen from Capital University of Medical Sciences .between September 2002 and January 2003.① Forehead-mamillary process of brain asmittance of the subjects in each grouP was given lead detection with ENG bilateral admittance rbeo-encephalogram automatic detection.②The wavelet pack technology was used to extract different frequencies of entropy feature of brain admittance to perform comparison and statistical analysis. RESULTS： Thirty patients respectively in the two groups and 30 healthy persons in normal control group entered the stage of result analysis. ① 2- grade decomposition of wavelet pack showed that the entroy value was both lower in hypertension group and arteriosclerosis group than in normal control group. There was significant difference of T （2,1）entroy feature value between hypertension group and arteriosclerosis group （303.61±90.48, 368.09±14.20,P=0.012）. ② 3-grade decomposition showed that there was significant difference of entroy feature value at middle-frequency between hypertension group and normal control group T（3,1）,T（3,2） and T（3,4） （P 〈 0.05）; The entroy feature in the arteriosclerosis group all had significant meaning. CONCLUSION： There is difference of entroy feature value extracted from brain admittance with wavelet packet technique between normal control group and abnormal group. Moreover, the difference at middle-frequency is significant. Decomposition of brain admittance and study of cerebral function state with wavelet package analysis technology will provide basis for further study such as signal classification.

关 键 词：脑导纳信号 熵 信号处理 计算机辅助 动脉硬化 

分 类 号：R319[医药卫生—基础医学]