Siyah Mansoory Meysam, Oghabian Mohammad Ali, Jafari Amir Homayoun, Shahbabaie Alireza
Department of Medical Physics & Biomedical Engineering, School of Medicine, Tehran University of Medical Sciences, Tehran, Iran.
Department of Neuro-Imaging and Analysis, Research Center for Molecular and Cellular Imaging, Tehran University of Medical Sciences, Tehran, Iran.
Basic Clin Neurosci. 2017 Sep-Oct;8(5):371-385. doi: 10.18869/nirp.bcn.8.5.371.
Graph theoretical analysis of functional Magnetic Resonance Imaging (fMRI) data has provided new measures of mapping human brain in vivo. Of all methods to measure the functional connectivity between regions, Linear Correlation (LC) calculation of activity time series of the brain regions as a linear measure is considered the most ubiquitous one. The strength of the dependence obligatory for graph construction and analysis is consistently underestimated by LC, because not all the bivariate distributions, but only the marginals are Gaussian. In a number of studies, Mutual Information (MI) has been employed, as a similarity measure between each two time series of the brain regions, a pure nonlinear measure. Owing to the complex fractal organization of the brain indicating self-similarity, more information on the brain can be revealed by fMRI Fractal Dimension (FD) analysis.
In the present paper, Box-Counting Fractal Dimension (BCFD) is introduced for graph theoretical analysis of fMRI data in 17 methamphetamine drug users and 18 normal controls. Then, BCFD performance was evaluated compared to those of LC and MI methods. Moreover, the global topological graph properties of the brain networks inclusive of global efficiency, clustering coefficient and characteristic path length in addict subjects were investigated too.
Compared to normal subjects by using statistical tests (P<0.05), topological graph properties were postulated to be disrupted significantly during the resting-state fMRI.
Based on the results, analyzing the graph topological properties (representing the brain networks) based on BCFD is a more reliable method than LC and MI.
功能磁共振成像（fMRI）数据的图论分析为体内人脑映射提供了新的测量方法。在所有测量区域间功能连接性的方法中，作为一种线性测量方法，计算脑区活动时间序列的线性相关性（LC）被认为是最常用的方法。由于并非所有双变量分布，而仅是边缘分布为高斯分布，因此LC始终低估了构建和分析图所需的依赖强度。在许多研究中，互信息（MI）已被用作脑区每两个时间序列之间的相似性度量，这是一种纯非线性度量。由于大脑复杂的分形组织表明其具有自相似性，通过fMRI分形维数（FD）分析可以揭示更多关于大脑的信息。
在本文中，引入了盒计数分形维数（BCFD），用于对17名甲基苯丙胺吸毒者和18名正常对照者的fMRI数据进行图论分析。然后，将BCFD的性能与LC和MI方法的性能进行了比较。此外，还研究了成瘾受试者脑网络的全局拓扑图属性，包括全局效率、聚类系数和特征路径长度。
通过统计检验（P<0.05）与正常受试者相比，静息态fMRI期间拓扑图属性被假定为显著破坏。
基于这些结果，基于BCFD分析图拓扑属性（代表脑网络）是一种比LC和MI更可靠的方法。
