Correlation thresholding of segmented fMRI time series baseline data for extraction of the node network structure in the human brain

Abstract

Functional magnetic resonance imaging (fMRI) is a technology that has greatly improved neuroscience research due to its noninvasive, relatively high-speed, and highly sensitive method to map human brain activity. This method allows the acquisition of thousands of measures in a single scan which can be used to detect abnormalities, diagnose conditions, understand brain disorders, and many more. Despite an abundance of data, fMRI scans are difficult to interpret due to variability in cognition of individuals and the experimental conditions that they underwent. Specific data pre-processing methods and statistical analysis techniques are being explored and reproduced to understand brain function. In this study, we aim to analyze brain network topology using the graph theoretical approach. In this regard, the brain is considered as a network of nodes composed of the cubic volume (voxel) where the MRI data is collected and edges are connections between these voxels according to how the activities are correlated between them. Using this method, the global network parameters were calculated and its relationship with thresholding is observed.