线性映射函数对无损颅内压估计的影响  

Linear mapping function to the influence of non-invasive ICP assessment

作　　者：吴少智[1,2] 吴跃[1] 徐鹏[3,2] 胡晓[2] 

机构地区：[1]电子科技大学计算机学院,成都611731 [2]加州大学洛杉矶分校 [3]电子科技大学神经信息教育部重点实验室,成都610054

出　　处：《电子测量与仪器学报》2010年第8期716-721,共6页Journal of Electronic Measurement and Instrumentation

基　　金：国家863计划(编号:2007AA01Z443)资助项目;博士点基金(编号:20070614008)资助项目;美国NINDSR21(编号:NS055998;NS055045;NS059797);R01(编号:NS054881)资助项目

摘　　要：提出了一种基于截断奇异值分解的线性映射函数颅内压无损估计方法。在基于时间序列的数据挖掘框架基础上,分别推导了基于总体最小二乘法的线性映射函数和基于截断奇异值分解法的线性映射函数,通过实验仿真比较证明:不同的线性映射函数对颅内压的预测确实有较为明显的影响,在所采用的方法中,基于截断奇异值分解法的线性映射函数效果明显优于基于总体最小二乘法的线性映射函数。因此,在该数据挖掘框架中,选择合适的线性映射函数对无损颅内压预测精确性的提高有着非常重要的意义。A linear mapping function based on truncated singular value decomposition for noninvasive estimation of intracranial pressure is presented.It was deduced to the linear of mapping function based on total least squares and truncated singular value decomposition,respectively,according to the time series data mining framework.Experimental simulation results proved： different linear mapping function has obvious effects to the non-invasive intracranial pressure estimation,the truncated singular value decomposition based mapping function show better intracranial pressure estimation performance with smaller intracranial pressure errors than the total least squares based mapping function.Therefore,select the appropriate mapping function of the non-destructive prediction accuracy increased intracranial pressure has very important significance in the time series data mining framework.

关 键 词：数据挖掘框架 线性映射函数 总体最小二乘法 截断奇异值分解法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]