Depressed MEG causality analysis based on polynomial kernel Granger causality

Abstract

In this study, we employ the Granger causality of a polynomial kernel to identify the coupling causality of depressed magnetoencephalography (MEG). We collect MEG under positive, neutral, and negative emotional stimuli and focus on the β-band activities. According to test results, depressed people display stronger left–right symmetrical interconnection in their prefrontal and occipital lobes under nonpositive stimuli(namely neutral and negative stimuli), indicating that they are more sensitive to nonpositive stimuli. The intensity of the right occipital information flow is higher in depressed people. We also see the Granger causality index increased in the occipital–frontal areas of depressed patients under negative stimuli. In general, detecting the polynomial kernel Granger causality of the MEG can effectively characterize the strength of the interconnected brain regions in depressed patients, which can be used as a clinical diagnosis aid.