基于子波变换阿尔茨海默病脑电信号多尺度定量特征的研究

Multi-scale analysis of EEG signals in patients with Alzheimer disease through wavelet transformation

目的：研究阿尔茨海默病（AlzheimerDisease，AD）脑电信号的多尺度定量特征和相位平均波形。方法：采集32例重度AD患者，30例轻度AD患者和30例正常对照的清醒安静闭目状态下的脑电信号，进行Gauss连续子波变换，提取脑电信号的时频分布特征和多尺度功率谱分布特征；应用条件采样和相位平均的方法提取脑电信号分尺度相位平均波形。结果：AD脑电信号的时频结构特征表现为尺度单一，节律性活动紊乱，而正常对照脑电信号尺度结构丰富，在0．1Hz、1Hz和10Hz频带上形成稳定的节律性活动。AD患者脑电信号的多尺度功率谱分布特征表现为在1Hz附近出现窄而高的功率峰，而正常对照老年人脑电信号表现为在0．1Hz、1Hz和10Hz附近出现三个宽而低的功率峰。多尺度相位平均波形的结果显示，不同导联脑电信号第9尺度（频率中心10Hz）的相位平均波形的波长在重度AD组、轻度AD组和正常对照组三组之间比较存在显著差异（P〈0．01），组间两两比较也存在显著差异（P〈0．05）。不同导联脑电信号第9尺度的相位平均波形的波长与简易智能精神状态量表（MMSE）评分之间存在负相关（P〈0．01），说明这一参数与病情严重程度相关。结论：子波分析适用于痴呆病人脑电信号的定量分析，研究表明脑电信号的时频结构、多尺度功率谱分布和第9尺度相位平均波形的波长可以作为AD诊断和评估的定量电生理指标。

Objective： To explorre multi-scale qualitative characteristics of EEG in patients with Alzheimer disease（AD）through wavelet transform. Methods：The EEG records of 32 severe patients with AD,30 mild and 30 normal controls were analyzed through wavelet transform. Characteristics of timefrequency structures of AD patients and normal controls were captured. Wavelet power spectrum was in- troduced to evidence multi-scale distribution index of EEG signals. Multi-scale phase averaged waveforms were extracted by wavelet coefficient index using conditional phase averaging technique. Results： Charac- teristics of time-frequency structures of AD was simple and in disorder, compared with rhythmic activi- ties at 0.1 Hz,1 Hz,10 Hz and co-related with each other in normal. The narrow-band power spectrum with single peak at 1 Hz was typical characteristic EEG in patients with AD, while normal controls always presented wideband power spectrum with three typical peaks at 0.1 Hz, 1 Hz, 10 Hz . The wavelength of phase averaged waveform of scale 9（10 Hz） at all leaders in severe AD group, mild AD group and con- trol group were different （P^0.01）. The wavelength of phase averaged waveform of scale 9 at all lead- ers was correlated with the score of MMSE in AD patients The parameter of frequency in EEG,was relat- ed of the severeness of the disease. Conclusion：Wavelet transform can reveal significant aspects of EEG in patients with AD. Wavelet power spectrum and the wavelength of phase averaged waveform at scale 9 are useful variables of quantity in the diagnosis and evaluation of AD.