HHT在血细胞特征提取中的应用  

Application in blood feature extraction with Hilbert-Huang transform

作　　者：罗亮[1] 龙伟[1] 李富贵[1] 詹从来 

机构地区：[1]南昌大学信息工程学院,江西南昌330031

出　　处：《电子技术应用》2015年第7期154-157,共4页Application of Electronic Technique

基　　金：国家自然科学基金资助项目(61261011)

摘　　要：提出一种基于Hilbert-Huang变换(HHT)的特征分析方法,该方法将血细胞信号进行经验模态分解和Hilbert变换,提取信号的平均强度、频谱质心和能量贡献率作为频域特征,与信号的时域特征结合,最终完成血细胞特征向量对脉冲信号的统计和识别。仿真实验中,使用HHT的识别算法正确率由模拟电路法的72.33%提高至94.33%;而使用该算法的血细胞分析仪与奥菲MYTHIC 18的可比性合格率达到98.5%,分类相关性系数都在94%以上。实验结果表明,该方法能提高仪器的计数正确度和分类准确性。This paper proposes a method that using Hilbert-Huang transform （HHT） to analyze the characteristic of blood cell signal. This method makes empirical mode decomposition and the Hilbert transform for blood cell signal. Extracted the signal average strength, the center of mass and energy contribution of spectrum as a frequency domain characteristics, and combined with time domain characteristics of the signal, it finally completes the recognition of the pulse signal with blood cell＇s eigenvector. The accuracy of using HHT identification algorithm is 94.33% while analog circuit method is 72.33%. The comparability is achieved by 98.5% and all of the correlation coefficient are more than 94% compared with MYTHIC 18. According to experimental results, this method can effectively improve the counting accuracy and classification accuracy of hematology analyzer.

关 键 词：血细胞信号 特征提取 经验模态分解 频域特征 HILBERT-HUANG变换 细胞分类 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]