基于希尔伯特-黄变换的白细胞信号分析

发布时间：2018-05-04 21:45

本文选题：白细胞 + 希尔伯特-黄变换　；参考：《中国生物医学工程学报》2014年01期

【摘要】：白细胞信号(WBS)具有脉冲形状多样和时频特性各异的特点,目前临床上用细胞信号脉冲计数的方法来分析WBS通常难以反映其所携带的丰富的生理和病理信息,并直接影响到细胞分类问题。针对这一问题,探索能自适应地分解非线性非平稳信号的希尔伯特-黄变换算法在WBC时频分析和分类中的应用效果。通过对血细胞中的WBC进行HHT变换,获取WBC的本征模态函数(IMF)分量、分量的Hilbert边际谱以及信号Hilbert谱;利用瞬时频率、瞬时幅值等进行计算提取健康人与患者的WBS平均强度、谱质心以及能量贡献率等特征作对比分析,根据其时频特征分布规律构建用于分类实验的特征向量;采用支持向量机(SVM)分类器,对58名健康人和60名患者的白细胞实验样本进行分类实验。结果表明,该方法提取的健康人和患者的WBS分量谱质心分布、平均强度值以及能量贡献率具有较好的区分度,分类正确率到达了94.83%。HHT方法能有效提取WBS特征,可辅助临床WBS的处理和分析。
[Abstract]:Leucocyte signal (WBS) has the characteristics of different pulse shape and different time frequency characteristics. At present, it is difficult to analyze the rich physiological and pathological information of WBS by the method of cell signal pulse counting, and directly affect the cell classification problem. The application effect of Hilbert - yellowish conversion of stationary signal in the time-frequency analysis and classification of WBC. By HHT transformation of WBC in the blood cells, the eigenmode function (IMF) component of WBC, the Hilbert marginal spectrum of the component and the Hilbert spectrum of the signal are obtained, and the WBS of healthy persons and patients is extracted by using instantaneous frequency and instantaneous amplitude. The average intensity, the spectral centroid and the energy contribution rate were compared and analyzed. According to the time frequency characteristic distribution, the characteristic vectors used for the classification experiment were constructed, and the support vector machine (SVM) classifier was used to classify the white cell experiment samples of 58 healthy people and 60 patients. The results showed that the healthy people and the patients were extracted by the method. The WBS component centroid distribution, the average intensity value and the energy contribution rate have good regional diversity. The correct classification rate reaches the 94.83%.HHT method and can effectively extract the WBS features, which can assist the treatment and analysis of the clinical WBS.

【作者单位】：南昌大学信息工程学院;
【基金】：国家自然科学基金(61261011)
【分类号】：R318.04

【相似文献】