基于SBF算法的HRV信号信息熵分析  

Information Entropy Analysis of HRV Signals Based on SBF Algorithm

作　　者：刘哲汝 张佃中[1] 黄立清[2] 

机构地区：[1]中南大学数学与统计学院,湖南长沙410083 [2]中南大学粉末冶金国家重点实验室,湖南长沙410083

出　　处：《中国医学物理学杂志》2014年第1期4652-4655,共4页Chinese Journal of Medical Physics

基　　金：国家自然科学基金(61271355);中南大学前沿研究计划(2010QZZD015)资助

摘　　要：目的:HRV信号是发生在非均匀间隔时间点上的RR序列,其传统的频域分析方法为功率谱分析,但对于非均匀采样的HRV信号,快速傅里叶变换(FFT)并不适用,同时HRV信号是具有混沌性和非平稳性的信号,功率谱也不善于表现HRV信号的非平稳性质。SBF(Similar Basis Function)算法是相对于FFT的另一种傅里叶积分估计方法,适用于均匀与非均匀采样信号,片段谱是基于SBF算法定义的表示信号能量分布的参数,相对于功率谱其主要的优点是能表现谱随时间的变化,对于处理非平稳信号也有一定的优势。因此本文探究用片段谱信息熵作为HRV信号的参数指标,分析HRV信号在不同频段能量分布复杂度随年龄的变化。方法:本文以20名年轻(21~34)岁与20名年老(68~81)岁二组健康人的HRV信号为实验数据,用SBF算法计算出二组人的片段谱,再算出多个不同频率段上的信息熵,同时用全频段分割法计算两组人的信息熵。结果:在0.003 Hz^0.04 Hz和0.04 Hz^0.15 Hz频段内,年轻组的片段谱熵明显大于老年组(p<0.001,p<0.01);由全频段分割法也得到类似的结果。结论:因而基于SBF算法的片段谱熵是分析HRV信号的有效指标。Objective： HRV （heart rate variability） signals represent the succession of values of varying intervals between heart beats occurring at non-equidistant sampling times,and the conventionary frequency domain measure is power spectrum of HRV in which fast Fourier transform （FFT） is not suited to. Besides, HRV has properties of chaos of non-stationary process, however, the power spectrum is defined as a frequency measure of a stationary random process. SBF （similar basis function） algorithm which is capable to deal with both uniform and non-uniform sampling signals, is a method to estimate Fourier integrals corresponding to FFT; FS （fragmentary spectrum） is a descriptor of power distribution of the signal based on SBF, the main advantage of FS is its ability to measure how spectral estimates evolve over time compared with the power spectrum. The work in this paper is to explore to use FSIE （fragmentary spectral information entropy） as a parameter to analyze how the energy distribution complexity of the HRV signal evolves with ages in different frequency segments. Methods： HRV signals are obtained from the young （21-34） and the elderly （68-81） groups of 40 healthy subjects, and Information entropies are calculated through fragmentary spectra by SBF algorithm on multiple various segments of frequency. Results：FSIE within 0.003-0.04Hz and 0.04-0.15Hz are obviously greater in young group than the healthy elderly group （p〈0.001,p〈0.01）.Similar results could be obtained by dividing up the whole frequency segment. Conclusions： FSIE of various frequency＇ segments is an available index to analyze HRV signals.

关 键 词：SBF算法 片段谱 心率变异性 非均匀采样 信息熵 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]