摘要
传统计算功能性连接(FC)的方法较少关注对应的频率信息,且所得聚类簇的生理意义存在争议,提出一种小波相干聚类方法。首先,采用小波相干研究FC的频率、相位差和时间差信息;其次,定义四种类型以明确聚类簇的生理意义;最后,采用K-means和欧氏距离判决相结合的方法寻找MCI和正常对照(NC)的簇间差异性。实验结果表明,type2中最大频率约0.11 Hz,最大时间差约46.33 ms,type1中最大频率约0.08 Hz,最大时间差约67.01 ms,两者间频带、相位差和时间差异度分别约为0.02 Hz、0.49 rad和20 ms。分析表明,在较大相干值时MCI则更多表现出小的相位差和时间差,且发生在较高频带slow4-3和slow4-4上;NC被试更多表现出大的相位差和时间差,且发生在较低频带slow4-1和slow4-2上。
Traditional methods for calculating functional connectivity(FC)pay less attention to the corresponding frequency information,and the physiological significance of the obtained clusters is controversial.Therefore,this paper proposed a wavelet coherent clustering method.Firstly,it used wavelet coherence to study the frequency,phase synchronization and time synchronization information of FC.Secondly,it defined four types to clarify the physiological meaning of clusters.Finally,it used the method of combining K-means and Euclidean distance judgment to find differences of clusters between MCI and normal controls(NC).The experimental results show that the maximum frequency in type2 is about 0.11 Hz,the maximum time synchronization is about 46.33 ms,the maximum frequency in type1 is about 0.08 Hz,and the maximum time synchronization is about 67.01 ms.The frequency band,phase synchronization and time synchronization between the two are about 0.02 Hz and 0.49 rad and 20 ms respectively.The analysis show that MCI subjects have more small phase synchronization and time synchronization when the coherence value is larger,which occurrs in the higher frequency bands slow4-3 and slow4-4;NC subjects have more large phase synchronization and time synchronization,which occurs in the lower frequency bands slow4-1 and slow4-2.
作者
保涵
吴海锋
晏洁
曾玉
Bao Han;Wu Haifeng;Yan Jie;Zeng Yu(School of Electrical&Information Engineering,Yunnan Minzu University,Kunming 650504,China;Intelligent Sensor Network&Information System&Technology Innovation Team in University of Yunnan Province,Yunnan Minzu University,Kunming 650504,China)
出处
《计算机应用研究》
CSCD
北大核心
2023年第2期526-532,共7页
Application Research of Computers
基金
国家自然科学基金资助项目(62161052)。
关键词
静息态功能核磁共振成像
动态功能性连接
相位同步
时间同步
聚类
rest-state functional magnetic resonance imaging
dynamic functional connectivity
phase synchronization
time synchronization
clustering
