Multi-Scale Analysis of Brain Functional Networks in Schizophrenia Based on Spectral Graph Wavelets

Schizophrenia is a serious mental illness whose pathological mechanism is still unknown. In this paper, we study brain functional networks of schizophrenic patients from a new perspective of graph signal processing. We use fMRI data of 71 schizophrenic patients and 74 healthy subjects to represent their graph signals and calculate their Laplace matrices by Pearson correlation coefficients. Based on the maximum Laplace eigenvalue, we construct spectral graph wavelet and decompose graph signals into five frequency scales. We compare the energy distribution of brain functional network in schizophrenic patients and healthy subjects. We find that, compared with healthy subjects, the energy distribution of brain functional networks in schizophrenic patients is significantly different at different scales, especially abnormal activities in brain regions: Cortex Left, Parahippocampal Gyrus and Anterior Division Right.