Analysis Of Power Spectrum Functional Brain Networks In Major Depressive Disorders Based On EEG Data

Major depressive disorder(MDD)is a affective disorder characterized by persistent,reduplicated feelings of sadness,guilt,and worthlessness.At present,the understanding of the pathogenesis of MDD is not deep enough,and the diagnosis and treatment methods are still not perfect.The study of MDD is still an important scientific research and clinical topic.Although the findings of graph theory research on the functional brain networks of MDD patients are not identical,the concept of network randomization has a strong convergence trend.EEG is an overall reflection of a large number of neural activities in the scalp,contains abundant physiological and pathological information,and has the characteristics of easy acquisition and real-time monitoring.In this study we aimed to further investigate the characteristics of the randomized functional brain networks in MDD patients by examining resting-state scalp-EEG data.Based on the methods of independent component analysis(ICA),fast Fourier transform(FFT)and graph theoretic analysis,the abnormalities in the power spectrum functional brain networks were compared between 13 MDD patients and 13 matched healthy controls(HCs).Nonparametric permutation tests were performed to explore the between-group differences in multiple network metrics.The Pearson correlation coefficients were calculated to measure the linear relationships between the clinical symptom and network metrics as well as the subjects' ages.In the alpha band,compared with the HCs,the MDD patients showed significant randomization of global network metrics,characterized by greater global efficiency,but lower connectivity strength,clustering coefficient,characteristic path length,and local efficiency.From the features of small-worldness,the functional brain networks of HCs and MDD patients all exhibited an apparent small-world architecture characterized by almost identical path length but more local clustering.In addition,the network efficiency analysis also demonstrated economic small-world configurations in all studied networks,as characterized by approximately equivalent parallel information processing capability but higher fault tolerance.However,compared with HCs,MDD patients had a significantly reduced normalized characteristic path length but increased normalized global efficiency.These results suggest that MDD patients and HCs had different small-world architectures for the functional brain networks.For MDD patients,the best-fitting model of degree distribution was an exponential truncated power law distribution.However,the probability distribution of node degree was somewhat less heterogeneous and less fat-tailed than that of HCs.Results of this study reveal that the randomized functional brain networks in MDD patients have greater resilience than those in the HCs whether attacked randomly or targeted,which might be a protective mechanism to avoid fast deterioration of the integrity of MDDs' brain networks under pathological attack.Essentially,the increase in resilience caused by randomization of the network is closely related to the abnormal reduction in the larger degree hub nodes.This study also reveals that the enhanced randomization and resilience in the brain networks of MDD patients have a reduced level of rich-club coefficient.Since the rich-club hubs are the part of the hub nodes that are more closely connected to each other,the reduced hub nodes in the brain networks of MDD patients may affect the density of connections among rich-club hubs.