一种基于复数主成分分析的磁共振波谱量化预处理方法  

A Preprocessing Method for Quantitation of Magnetic Resonance Spectroscopy Based on Complex Principal Component Analysis

作　　者：曾卫明[1] 梁晓云[1] 罗立民[1] 

机构地区：[1]东南大学生物医学工程系,南京210096

出　　处：《电子与信息学报》2005年第1期51-55,共5页Journal of Electronics & Information Technology

摘　　要：该文提出了一种应用复数主成分分析方法(CPCA)来获取波谱基木波型,自动校正频率和衰减系数偏移的方法。该方法首先对频率进行校正,然后拓宽CPCA的应用范围,把它使用到衰减系数偏移校正领域。文中的实现结果证明:该方法较好地实现了频率和衰减系数的校正,在同实域PCA方法性能比较上,前者的计算速度和估计精度明显占优。This paper proposes a method that can automatically obtain common lineshape in the spectrum data set based on Complex Principal Component Analysis(CPCA), which can correct frequency shifts and damping factor shifts effectively. This method aligns frequency at first, then extends CPCA to the correction of damping factor shifts in a single resonant peak across spectra. Through this method, frequency and damping factor alignment are successfully achieved. In terms of computation time and estimation precision this method is significantly better than that of PCA.

关 键 词：复数主成分分析 磁共振波谱 频率校正 衰减系数校正 

分 类 号：R8[医药卫生—临床医学]