脑电信号的标度分析及其在睡眠状态区分中的应用

• 作 者: 马千里 卞春华 王俊
• 机 构: 南京大学电子科学与工程学院,近代声学教育部重点实验室,生物医学电子工程研究所,南京邮电大学地理与生物信息学院,图像处理与图像通信江苏省重点实验室,
• 出 处: 物理学报
• 发表日期: 2010年07期
• 关键词: 睡眠状态 脑电信号 标度 多重分形 睡眠分期 波动分析 定量参数 生理病理 fluctuation 时间序列

脑电信号具有长程幂律相关性及多重分形的标度特性,并随生理病理状态改变.本文首次针对睡眠脑电信号应用单重分形去趋势波动分析(detrended fluctuation analysis,简记为DFA)方法与多重分形奇异谱对睡眠脑电信号的标度特征进行系统的对比研究.发现DFA标度指数α对于不同导联和样本组间的差异较为敏感,随睡眠状态的变化不规律;而多重分形奇异强度区间Δα随睡眠状态的变化更为规律,睡眠Ⅰ期至Ⅳ期不断增大,并且导联间差异和样本组间差异均较小.多重分形Δα参数更适合作为判定睡眠状态的定量参数.

EEG signals have long-range power-law correlation and multifractal scaling characteristics, and they are changed with the pathophysiological and pathological conditions.In this paper, we first apply the detrended fluctuation analysis (DFA) method to sleep-induced brain electrical signals And multifractal singular spectrum of sleep EEG scaling characteristics of the system were compared.DFA scale index α found for different lead and sample groups are more sensitive to the variation with the sleep state of the irregular change and the multifractal The singular intensity interval Δα was more regular with the change of sleep state, and the sleep stages Ⅰ to Ⅳ increased continuously, and the difference between lead and sample groups was small. The multifractal Δα parameter was more suitable as a quantitative parameter to determine the sleep state .