基于多尺度排序递归图的糖尿病轻度认知障碍脑电信号分析

Multi-Scale Recurrence Plot Based EEG Signal Analysis of Diabetes Mellitus and Mild Cognitive Impairment

研究糖尿病轻度认知障碍(MCI)脑电信号多尺度的确定性特性,有利于其早期预防和诊断。基于多尺度下的排序递归图,对15通道闭眼静息态的脑电信号进行多尺度排序递归图确定性分析。4组被试分别为14例糖尿病轻度认知障碍(DM-MCI)、11例糖尿病认知功能正常(DM-nMCI)的患者以及16例无糖尿病轻度认知障碍(n DM-MCI)、10例无糖尿病认知功能正常(n DM-nMCI)的患者。利用单因素方差分析(one-way ANOVA)对4组被试脑电信号的多尺度确定性值进行统计分析,利用皮尔森相关方法研究多尺度确定性与认知功能之间的相关性。结果显示,小尺度1≤s≤4时,n DM-MCI组和n DM-nMCI组确定性DET值明显高于DM-MCI组和DM-nMCI组的DET值;大尺度12≤s≤15时,4组DET值差异性小,且DM-MCI组所有电极不同尺度下的确定性值均要高于DMn MCI;轻度认知障碍患者在C3(DM-MCI:0.83±0.09,n DM-MCI:0.86±0.09)和C4(DM-MCI:0.85±0.08,n DMMCI:0.88±0.08,)电极的小尺度下确定性值显著增高,是认知障碍的主要特征,且与Boston(P=0.033),FAQ(P=0.026),WAIS(P=0.037),MoCA(P=0.039,P=0.017)等多项认知功能显著相关;糖尿病患者在Fp1(DMMCI:0.80±0.09,DM-nMCI:0.75±0.07)、Fp2(DM-MCI:0.80±0.09,DM-nMCI:0.75±0.01)和Oz(DM-MCI:0.76±0.06,DM-nMCI:0.73±0.07)小尺度确定性和Fp2(DM-MCI:0.96±0.03,DM-nMCI:0.94±0.01)大尺度确定性显著降低,是糖尿病的重要特征。基于多尺度排序递归图的确定性可用于分析与认知功能下降和糖尿病相关的脑电特征。

Studying the EEG characteristics in mild cognitive impairment(MCI)of diabetes is helpful to the early prevention and diagnosis of the disease.Based on the multi-scale recurrence plot.We analyzed the deterministic characteristics of 15-electrode eye-closed resting state EEG of 14 patients with mild diabetes mellitus MCI(DM-MCI)and 11 patients with normal cognitive function of diabetes mellitus(DM-nMCI),and16 patients with mild cognitive impairment without diabetes mellitus(n DM-MCI)and 10 patients with normal cognitive function without diabetes mellitus(nDM-nMCI).Deterministic values between the four groups were statistically analyzed using one-way ANOVA,and the correlations between deterministic values and cognitive function of each electrode were calculated by Pearson correlation method.The results showed that at small scale 1≤s≤4,the values of DET in patients with nDM-MCI and n DM-nMCI were significantly higher than those with DM-MCI and DM-nMCI;At large scale 12≤s≤15,four groups of DET values tend to be consistent.The deterministic values of all electrodes at different scales in the DM-MCI group were higher DMnMCI group.The deterministic values of MCI at C3(DM-MCI:0.83±0.09,nDM-MCI:0.86±0.09)and C4(DM-MCI:0.85±0.08,nDM-MCI:0.88±0.08)electrodes increased significantly at different scales,which were important characteristics in distinguishing MCI and their correlations between DET and neuropsychological test scores Boston(P=0.033),FAQ(P=0.026),WAIS(P=0.037),and MoCA(P=0.039,P=0.017)was significant.The DET values of Fp1(DM-MCI:0.80±0.09,DM-nMCI:0.75±0.07),Fp2(DM-MCI:0.80±0.09,DM-nMCI:0.75±0.01)and Oz(DM-MCI:0.76±0.06,DM-nMCI:0.73±0.07)at small scale,and Fp2(DM-MCI:0.96±0.03,DM-nMCI:0.94±0.01)at big scale increased significantly,which are the important characteristics in distinguishing DM.It was suggested that the determinacy based on multiscale order recurrence plot was the EEG feature associated with cognitive decline and DM.