基于稳态视觉诱发电位和精神分裂症脑磁信号的分析与识别研究

发布时间：2018-04-08 19:12

本文选题：脑信号　切入点：精神分裂症　出处：《南京邮电大学》2017年硕士论文

【摘要】：脑电/磁信号是能够反映大脑不同生理状态的复杂生物信号,常用于癫痫、老年痴呆症、精神分裂症等疾病的诊断和监测,病理脑电/磁信号的分析有助于进一步地了解脑疾病的基础发生机制,为脑疾病的的临床诊断提供参考依据,为患者的康复带来希望,论文从精神分裂症脑磁信号以及稳态视觉诱发电位脑电信号两方面进行了研究。论文首先对脑电/磁信号的分析方法进行了介绍,重点介绍了特征提取和模式分类方法,对这些方法的原理和特性进行了较为详细的阐述。论文提出了一种基于多维复杂度的脑磁信号分析方法。通过提取精神分裂症脑磁信号的AR模型系数、频带能量、近似熵和Lempel-Ziv复杂度作为特征,运用距离准则和增L减R算法进行通道筛选,再运用BP神经网络和支持向量机对精神分裂症和正常人的脑磁信号进行区分,分类正确率分别为96.25%和98.75%,实验表明该方法可以有效地区分精神分裂症患者和正常人。论文还运用遗传算法选择具有显著性差异的特征,BP神经网络和支持向量机的分类正确率分别为98.5%和99.75%,支持向量机可以获得更好的分类性能。最后,论文基于SSVEP设计脑电实验,采集多位被试的脑电信号,运用DFT、CCA以及MSI分析方法对采集的信号进行分析,DFT方法的频谱分析结果表明信号能量在目标刺激频率处最大,CCA和MSI方法分析过程中可以观察到信号在目标频率处的相关系数以及同步指数最大,无论是短时间窗还是长时间窗,CCA方法的识别结果要优于MSI方法和DFT方法,而MSI方法性能总体上要优于DFT方法,尤其是在数据长度较短时。
[Abstract]:EEG / magnetic signals are complex biological signals that can reflect different physiological states of the brain. They are often used in the diagnosis and monitoring of epilepsy, Alzheimer's disease, schizophrenia and other diseases.The analysis of pathological EEG / magnetic signals is helpful to further understand the underlying mechanism of brain disease, to provide a reference for the clinical diagnosis of brain disease, and to bring hope for the recovery of patients.In this paper, the EEG signal of schizophrenia and the steady state visual evoked potential (VEP) are studied.In this paper, the analysis methods of EEG / magnetic signals are introduced, especially the methods of feature extraction and pattern classification, and the principle and characteristics of these methods are described in detail.In this paper, a method based on multi-dimensional complexity for the analysis of brain magnetic signals is proposed.The AR model coefficients, frequency band energy, approximate entropy and Lempel-Ziv complexity of the brain magnetic signals of schizophrenia were extracted as the features. The distance criterion and the algorithm of increasing L minus R were used to screen the channels.Then BP neural network and support vector machine were used to distinguish the brain magnetic signals between schizophrenia and normal subjects. The classification accuracy was 96.25% and 98.75% respectively. The experiment shows that the method can effectively distinguish schizophrenia patients from normal people.The classification accuracy of BP neural network and support vector machine are 98.5% and 99.75, respectively. Support vector machine can obtain better classification performance.Finally, the experiment of EEG was designed based on SSVEP, and the EEG signals were collected.The spectrum analysis of the collected signals by DFT and MSI shows that the signal energy can be observed at the target frequency when the signal energy is maximum at the target frequency and the MSI method can be used to analyze the signal at the target frequency.Correlation coefficient and synchronization index are the largest,The results of MSI and DFT are better than that of MSI and DFT, and the performance of MSI is better than that of DFT, especially when the length of data is short.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R749.3;TN911.6

【参考文献】