一种基于稀疏表示的脑功能子网络构建方法

Method for constructing brain functional sub-networks based on sparse representation

针对根据大脑功能磁共振成像数据构建脑功能交互子网络的问题。将小波变换一致性定义为感兴趣区域(region of interest,ROI)之间功能交互模式的测度,通过稀疏表示提取每个ROI的特征,最后运用谱聚类的方法对其进行聚类,聚类结果将大脑划分为八个功能交互子网络。结果表明,这八个子网络分别对应着大脑的视觉、运动感知等主要功能系统,与当前的神经科学知识相一致,为脑功能网络研究提供了一种新的思路。

This paper focused on the construction of brain functional interaction sub-networks from functional megnetic reso- nance imaging （fMRI） data. First, it defined wavelet transform coherence （WTC） as the measure of functional interaction pat- tern between different ROIs （ region of interest）, extracted features of each ROI with sparse reperesentation afterwards, clus- tered 8 sub-networks using spectral clustering at last. The result shows these 8 sub-networks concide with brain＇ s functional systems such as vision, sensory-motor ere, and are consistent with current neuroscience knowledge. Therefore, the proposed method provides a novel train of thought in the research of brain functional networks.