基于fMRI数据的改进模糊聚类算法研究

Research on Improved Fuzzy Clustering Algorithm Based on fMRI Data

针对功能磁共振成像(functional Magnetic Resonance Imaging,f MRI)数据的特点,在聚类fMRI数据时,两个体素之间的距离通常并非是一种简单的空间距离,而是和它们之间的相关程度有关的距离测度.在双曲相关系数(Hyperbolic Correlation Coefficient,HCC)距离度量基础上发展了一种新的距离测度,并提出了改进模糊聚类算法的定义、算法评价标准和算法参数的优化方法,然后将改进后的模糊聚类算法分别在任务态和静息态f MRI数据上与基于HCC距离度量的模糊聚类算法进行比较,最后通过比较结果证明改进后的模糊聚类算法具有较好的聚类效果和收敛速度.

According to the characteristics of fMRI data,the distance between two voxels is usually not a simple spatial distance,but a distance measure is related to the correlation between them when clustering fMRI data.In this paper,a new distance measure based on HCC distance measure is developed,and the definition of the improved fuzzy clustering algorithm,the evaluation criteria of the algorithm and the optimization method of algorithm parameters are proposed.Then,the improved fuzzy clustering algorithm is used in task-related fMRI data and resting-state fMRI data respectively,and the experimental results are compared with those of fuzzy clustering algorithm based on HCC distance measure.Finally,the comparative results show that the improved fuzzy clustering algorithm has better clustering results and higher convergence speed than the fuzzy clustering algorithm based on HCC distance measure