基于复杂网络构建与分析技术的语音响度差异神经处理机制研究

相关分析能够找出研究现象之间的依存关系、相关方向以及相关程度,可以发现大数据集里隐藏的关联网络.本文面向语音响度变化认知问题,提出“差异度”的概念,利用相关分析构建大脑功能的复杂网络,探索深层的神经处理机制与脑认知新规律.提出一种短时窗分析方法,构建不同认知阶段的脑网络;基于不同刺激下节点度的拓扑特征,构建基于差异度的脑地形图,实现脑区之间数据关系的可视化表达和动态演化过程表达.结果发现,前额叶、右额颞区和右后颞区分别在听觉处理的早期、中期和晚期对声音响度变化具有显著响应.研究表明脑复杂网络构建与分析技术可以成为研究神经处理机制与认知规律的有效工具.

Research on complex network construction and analysisin speech loudness difference neural processing mechanism

The Correlation Analysis （CA） is capable to find out the dependence relationship, the correlation direction and thecorrelation degree between two objects, and can discover the correlation networks hidden in large data. Facing the problem ofspeech loudness perception study, the concept of ＂difference degree＂ is proposed and the CA is applied to build the complex brainfunction network, in order to explore the deep level neural processing mechanismand nouvelle brain cognitive principles. Ashortwindowbased analysis technology is proposed to construct a series of brain networks at different cognitive stages. Considering thetopological characteristics of the node degrees at different stimuli, a brain map can be constructed based on the ＂difference degree＂of nodes, as the result, the relationship among various brain areas is clearly visualized and the dynamic evolution process is preciselypresented. The experimental results revealed that the prefrontal, the right fronto-temporal and the right posterior temporal areasproduce great response to auditory loudness change separately during the early, middle and late auditory cognition stages. Our researchshows that the complex brain network construction and analysis technology will become an effective tool for the neural processingmechanismand cognitive principle studies.