以静息态脑电信号为基础,通过固有模态分解(empirical mode decomposition,EMD)算法对脑电信号进行信号去噪和特征值提取,通过支持向量机(support vector machine,SVM)算法对抑郁症患者和正常对照组人群的脑电特征值进行分类分析。通过系统化的数据采集试验,采集了20位抑郁症患者和25位健康对照组的静息态脑电信号;对静息态脑电信号进行信号的去噪和特征提取;采用SVM算法对抑郁症患者和正常人对照组脑电特征值进行二值分类,分类正确率达到93.3%。相较于传统的小波变换提取的特征值,分类准确率有明显的提高。 Based on resting EEG signals, the EEG signals were denoised and eigenvalues ​​extracted by using the empirical mode decomposition (EMD) algorithm. The support vector machine (SVM) EEG patients and normal control group were analyzed for EEG features. Through systematical data acquisition experiments, we collected resting EEG signals of 20 depressed patients and 25 healthy control subjects; carried out signal denoising and feature extraction of resting EEG signals; Patients with normal subjects and control group EEG binary classification, classification accuracy rate of 93.3%. Compared with the traditional eigenvalue extracted by wavelet transform, the classification accuracy is obviously improved.