计算机工程

• 人工智能及识别技术 • 上一篇    下一篇

EEG信号动态演化过程的研究

姜 慧,周 霆   

  1. (福州大学物理与信息工程学院,福州 350002)
  • 收稿日期:2012-04-24 出版日期:2013-09-15 发布日期:2013-09-13
  • 作者简介:姜 慧(1986-),女,硕士研究生,主研方向:模式识别;周 霆,副教授
  • 基金项目:
    福建省教育厅基金资助项目(JA08006);福州大学科研启动基金资助项目(022240)

Research on Dynamical Evolution Process of EEG Signal

JIANG Hui, ZHOU Ting   

  1. (College of Physics and Information Engineering, Fuzhou University, Fuzhou 350002, China)
  • Received:2012-04-24 Online:2013-09-15 Published:2013-09-13

摘要: 脑电图(EEG)信号的研究是诊断脑疾患的重要手段。以癫痫脑电为例,针对癫痫发作过程的复杂性,对其演化过程进行研究。利用本征正交分解(POD)对EEG信号实行特征压缩,选取能够反映EEG脑电病理特征的多个变量,通过改进的Fisher判别方法判别分解后的信号数据,以最终确定EEG信号动态演化过程的关键点。实验结果表明,将POD分解与Fisher判别方法相结合,不仅能减少数据分析的工作量,而且能够有效判别分析EEG信号动态演化过程。

关键词: 脑电图, 本征正交分解, Fisher判别, 特征压缩, 动态演化

Abstract: The research of Electroencephalogram(EEG) signals is an important means of diagnosis of brain disease. Taking EEG signals of epilepsy for example, for the complexity of the seizures, the evolution process is studied. It uses of the method of the Proper Orthogonal Decomposition(POD) to decompose and compress the EEG signals, chooses multiple variables to reflect electrical pathological characteristic of EEG brain, and uses the improved method of Fisher discrimination to classify the signal, determines the key points of the dynamic evolvement process of EEG signals. Experimental results show that combined with POD decomposition and Fisher discriminant method, it can not only reduce the workload of data analysis, and can effectively distinguish the EEG signal dynamic evolution process.

Key words: Electroencephalogram(EEG), Proper Orthogonal Decomposition(POD), Fisher discrimination, feature compression, dynamical evolution

中图分类号: