基于去趋势移动平均的抑郁症脑电图二元回归分析  

Detrended moving average-based binary regression analysis of electroencephalogram in depression

作　　者：陈剑 白登选 王琼 姚文坡 乙万义[2] 戴加飞 王俊[2] CHEN Jian;BAI Dengxuan;WANG Qiong;YAO Wenpo;YI Wanyi;DAI Jiafei;WANG Jun(School of Communications and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210009;School of Geographic and Biologic Information,Nanjing University of Posts and Telecommunications,Nanjing 210023;Department of Neurology,Jinling Hospital,Medical School of Nanjing University,Nanjing 210008)

机构地区：[1]南京邮电大学通信与信息工程学院,南京210009 [2]南京邮电大学地理与生物信息学院,南京210023 [3]南京大学医学院金陵医院神经内科,南京210008

出　　处：《北京生物医学工程》2023年第6期597-603,共7页Beijing Biomedical Engineering

摘　　要：目的本文运用基于去趋势移动平均(detrended moving average,DMA)的二元线性回归分析方法,对抑郁症患者组和健康对照组的脑电数据进行分析,探究抑郁症患者和健康人枕区对额区影响的差异性。方法基于DMA的二元回归模型不仅能够分析非线性和幂律相关的时间序列,还可以分析不同尺度下自变量和因变量间的依赖关系。通过该模型对抑郁症患者组和健康对照组的左、右枕区的对称导联和额区导联的脑电数据进行分析。同时本文使用原始的基于最小二乘法的二元线性回归模型对两组的脑电数据做相同处理,并对两种模型的实验结果进行对比。结果基于DMA的二元回归模型在尺度为264,自变量为左、右枕区导联,因变量为左额区导联时,抑郁症患者和健康人的回归系数存在显著性差异。而原始模型只有在极个别的以左、右枕区导联为自变量和左额区导联为因变量时存在显著性差异。存在显著性差异的回归系数均值β_(1)均大于0,回归系数β_(2)均小于0。结论基于DMA的二元回归模型不仅可以在不同尺度上描述大脑枕区和额区间的依赖性,还可用于分析抑郁症患者和健康人的脑电信号在左、右枕区对左额区影响的差异性。抑郁症患者的枕区对左额区的影响降低。Objective In this paper,the detrended moving average(DMA)-based binary linear regression analysis method is used to analyze the electroencephalogram(EEG)data of depressed patients and a healthy control group to explore the difference in the impact of occipital region on the frontal region between patients with depression and healthy people.Methods The DMA-based binary regression model can analyze nonlinear and power-law-related time series and the dependence of independent and dependent variables at different scales.The model analyzes the EEG data of left and right occipital symmetrical leads and frontal leads in patients with depression and healthy controls.At the same time,the original binary linear regression model based on the least square method is used to do the same processing on the EEG data of the two groups,and the experimental results of the two models are compared.Results When the scale of the DMA-based binary regression model is 264,the independent variable is left and right occipital leads and the dependent variable is left frontal leads,there are significant differences in the regression coefficients between depressed patients and healthy people.However,the original model has significant differences in only a few leads when the left and right occipital region leads are independent variables,and the left frontal region leads are dependent variables.Regression coefficient mean β_(1) with significant differences is greater than 0,and regression coefficient β_(2) is less than 0.Conclusions The DMA-based binary regression model can not only describe the dependence of the occipital and frontal regions of the brain at different scales but also analyze the difference of the effects of EEG signals in the left and right occipital regions on the left frontal regions between depressed patients and healthy people.In depressed patients,the occipital area has a reduced influence on the left frontal area.

关 键 词：脑电图 抑郁症 去趋势移动平均 不同尺度 二元线性回归分析 

分 类 号：R318.04[医药卫生—生物医学工程]