计算机工程与应用 ›› 2021, Vol. 57 ›› Issue (5): 245-250.DOI: 10.3778/j.issn.1002-8331.2008-0261

• 工程与应用 • 上一篇    下一篇

卷积神经网络及其分析在抑郁症判别中的应用

王凤琴,柯亨进   

  1. 1.湖北师范大学 物理与电子科学学院,湖北 黄石 435106
    2.武汉大学 计算机学院,武汉 435001
  • 出版日期:2021-03-01 发布日期:2021-03-02

Application of CNN and Its Analysis in Depression Identification

WANG Fengqin, KE Hengjin   

  1. 1.School of Physics and Electronic Science, Hubei Normal University, Huangshi, Hubei 435106, China
    2.School of Computer Science, Wuhan University, Wuhan 435001, China
  • Online:2021-03-01 Published:2021-03-02

摘要:

在线脑电分类能准确评估严重抑郁症患者的脑状态并及时跟踪其发展状态可以将其陷入危险和自杀的风险降为最低。然而,在无经验监督条件下,在线脑电分类应用面临更大的挑战:脑电数据往往具有弱信号、高噪声与非平稳特性;缺乏有效解耦脑疾病发作时脑区与神经网路的复杂关系。为此,设计一个以卷积神经网络为核心的云辅助在线脑电分类系统,该系统直接应用于原始脑电信号,无需进行预处理和特征提取,能精准、快速判别抑郁症状态。在公开数据集上进行抑郁症评估实验,对健康控制组和抑郁症对照组分类的准确率、敏感度和特异度分别为99.08%、98.77%和99.42%。另外,通过对神经网络进行定量解释,表明抑郁症病人的左右颞叶脑区与正常人存在明显差异。

关键词: 神经网络, 模型解释, 抑郁症, 脑电分类, 云计算

Abstract:

Online EEG classification can accurately assess the brain status of patients with Major Depression Disable(MDD) and track their development status in time, which can minimize the risk of falling into danger and suicide. However, it remains a grand research due to the embedded intensive noises and the intrinsic non-stationarity determined by the evolution of brain states, the lack of effective decoupling of the complex relationship between brain region and neural network during the attack of brain diseases. This study designs an online EEG classification system aided by cloud centering on a CNN. Experiments on depression evaluation has been performed against raw EEG without the need for preprocessing and feature extraction to distinguish Healthy & MDD. Results indicate that MDD can be identified with an accuracy, sensitivity, and specificity of 99.08%, 98.77% and 99.42%, respectively. Furthermore, the experiments on quantitative interpretation of CNN illustrate that there are significant differences between the left and right temporal lobes of depression patients and normal control group.

Key words: neural network, model interpretation, depression, EEG classification, cloud computation