基于稀疏表示的帕金森功能连接模式定位
Pattern localization of functional connectivity in Parkinson’s disease based on sparse representation
摘要点击 255  全文点击 122  投稿时间:2016-08-22  修订日期:2017-03-14
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2017.60629
  2017,34(6):843-848
中文关键词  模式定位  稀疏表示  多变量模式分析  功能连接
英文关键词  pattern localization  sparse representation  multivariate pattern analysis  functional connectivity
基金项目  国家自然科学基金;省自然科学基金
学科分类代码  
作者单位E-mail
陈勇斌 华南理工大学 mailcyb@163.com 
李远清 华南理工大学 auyqli@scut.edu.cn 
中文摘要
      在脑成像数据分析中, 基于稀疏表示的模式定位算法在群组水平分析中具有非常优秀的性能, 但在单个数据集的情况下结果还不尽如人意. 为此, 文中在先前研究的基础上提出了一种改进算法, 通过基于原始数据集生成多个派生数据集的方法, 来改善算法在单个数据集分析中的不足. 仿真结果表明改进后算法在性能上有显著的提高. 文章随后将该改进算法应用于帕金森病异常功能连接模式定位分析之中, 得到广泛分布于全脑的与该疾病相关的269个异常功能连接, 由此对算法的有效性进行了验证, 并可能有助于加强对与该疾病相关的病理生理机制的了解.
英文摘要
      In the analysis of brain imaging data, the sparse representation-based pattern localization algorithm has a very good performance at the group level data analysis. But at the single level, it’s performance is still disappointed. Therefore, in order to compensate for this deficiency, an improved algorithm based on previous research was proposed in this study. By generating multiple derived data sets from the original data and then performing pattern localization procedure, the improved algorithm has better performance compared to the original in simulation. Subsequently, the improved algorithm was applied to the analysis of localizing all abnormal brain functional connections in Parkinson’s disease. 269 abnormal connections were obtained and they were widely distributed throughout the entire brain. Thus, the effectiveness of the algorithm was verified and our findings may have the potential to advance the understanding of the neural mechanism of this disease.