Combining Graph Analysis and Recurrence Plot on fMRI data

Abstract

In this work we investigate on the nonlinear properties of the brain networks using Graph Analysis and Cross Recurrence Plot. The nonlinear dynamics of the brain is analyzed using time series coming from fMRI data. Two groups of human subjects, one affected by schizophrenia and the other of healthy controls, are imaged during the completion of a working memory task. To examine the spatio-temporal properties of the BOLD signal, nonlinear recurrence properties are extracted from the time series of the most relevant voxels, using the cross recurrence plots and the corresponding measures. Then, a graph is built using such measures as weights between different brain regions (the nodes). The purpose of the paper is to give a description of the most relevant functional areas activated during the task completion and to capture the differences between the groups. Results are promising, since the methodology is still to be fully developed and explored.