Detection Methods of Brain Activity under Visual Cognitive Tasks Based on Information Processing of EEG Signals

基于脑电信号信息处理的视觉认知任务下大脑活动检测方法

• 批准号: 09650477
• 负责人: MAEDA Makoto
• 金额: $ 2.24万
• 依托单位: Kyushu Institute of Technology
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (C)
• 财政年份: 1997
• 资助国家: 日本
• 起止时间: 1997 至 1999
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-09650477/
• 关键词: Electroencephalogram; Event Related Potential; Visual Evoked Response; Oddball Task; Identification; Signal Processing; Pattern Recognition; Human Interface; 選択反応実験; 確率システム

In this research project, we have performed studies on the development of detection method of brain activity under visual cognitive tasks based on information processing of electroencephalogram (EEG) signals and obtained following research results.1. EEG signal processing methodsWe have developed feature extraction and discrimination methods or EEG waves based on AR-model, Quasi ARMAX model and artificial neural networks. And it was also investigated that applied methods or wavelet analysis, blind separation analysis and event related de-synchronization analysis for EEG signals. An adaptive clustering method based on stochastic Newton method was developed and applied to EEG wave analysis. And it was confirmed that ERP (Event Related Potential) waves related to same kind of stimulus were belonged to some classes.2. Visual stimulating period and effect of eye blinkOptimal visual stimulating period were about 500msec for the discrimination or ERP signals, And eye blinks generated the noise component of EEG waves similar to ERP in shape. The effects of eye blink for the discrimination of EEG waves are under investigation.3. Discrimination of EEG waves under visual cognitive tasksThe effective electrode positions for the discrimination of ERP were Fz and Pz. Discrimination of stimuli (target or non-target) was tried based on EEG signal processing (AR model and statistical pattern recognition method). The accuracy based on regional EEG signals (spatial filtered signals) ws above 80%. These results suggest that human interface based on EEG signal processing would be feasible in near future.

本课题针对基于脑电信号信息处理的视觉认知任务下大脑活动检测方法的开发进行了研究，取得了以下研究成果： 1．脑电信号处理方法我们开发了基于AR模型、准ARMAX模型和人工神经网络的脑电波特征提取和判别方法。并探讨了脑电信号的小波分析、盲分离分析和事件相关去同步分析的应用方法。开发了一种基于随机牛顿法的自适应聚类方法，并将其应用于脑电波分析。并证实与同类刺激相关的ERP（事件相关电位）波属于某些类别。 2．视觉刺激周期和眨眼的影响对于ERP信号的辨别，最佳视觉刺激周期约为500msec，而眨眼产生的脑电波的噪声成分与ERP形状相似。眨眼对脑电波辨别的影响正在研究中。3．视觉认知任务下脑电波的辨别 ERP辨别的有效电极位置为Fz和Pz。尝试基于EEG信号处理（AR模型和统计模式识别方法）来区分刺激（目标或非目标）。基于区域脑电图信号（空间滤波信号）的准确度超过 80%。这些结果表明基于脑电图信号处理的人机界面在不久的将来将是可行的。

项目成果

H.Shiratsuchi, H.Gotanda, K.Inoue&K.Kumamaru: "A study on Initialization for BP Networks"Proc.of the 5th International Conference on Neural Information Processing. 1595-1599 (1998)

H.白土、H.五反田、K.井上

H.Gotanda, H.Shiratsuchi, K.Inoue&K.Kumamaru: "Necessary Conditions for Multilayer Nets Having Solutions and Convergent Superiority of Bipolar Nets"Proc.of the 11th IFAC Symposium on System Identification. 817-822 (1997)

H.五反田、H.白土、K.井上

井上 勝裕: "視覚認識時の脳波変動検出法"計測自動制御学会 第38回学術講演会予稿集. 509-510 (1999)

Katsuhiro Inoue：“视觉识别时的脑电图波动检测方法”第38届仪器与控制工程师学会学术会议论文集509-510（1999）。

熊丸耕介: "ウェーブレットネットワークを用いたハイブリット準ARMAXモデルの同定性能比較" 計測自動制御学会九州支部 第17回学術講演会予稿集. 241-242 (1998)

Kosuke Kumamaru：“使用小波网络的混合准ARMAX模型的识别性能比较”仪器与控制工程师学会九州分会第17届学术会议论文集241-242（1998）。

Makoto Maeda: "3-D Surface Recovery From Range Images by Using Multiresolution Wavelet Transform"Proc. of IEEE International Conference on Systems, Man, and Cybernetics. 3654-3659 (1997)

Makoto Maeda：“使用多分辨率小波变换从距离图像中恢复 3-D 表面”Proc。