Frequency-specific Network From Empirical Mode Decomposition In Major Depressive Disorder

Major depressive disorder(MDD),as the one of the common emotional disorder,is clinically manifested as constantly anhedonia,depressed mood and cognitive impairment.Some studies have used MRI to explore the abnormalities structure and function of brain in MDD patients,and it has been reported that MDD shows frequency dependent abnormal patterns in ALFF and Reho.However,in those studies,the frequency bands were divided by prior experience,and did not pay attention to the intrinsic information of BOLD signal.In this study,empirical mode decomposition(EMD),a data-driven frequency division method,was used to explore the frequency specific abnormal pattern of the triple network model and the whole brain network in MDD.This study consists of two parts:The first part investigates the frequency-based abnormal model of MDD patients’ cognitive core triple network model.Previous studies have founded that the abnormal patterns exist in default model network,salient network,and central executive network in MDD,but whether such abnormalities are frequency dependent is not yet clear.In this study,seven sub-networks from the triple network model was geted by the high-order independent component analysis,contains a DMN,ip DMN,sp DMN,lv CEN,rv CEN,d CEN and SN.EMD was used to explore the triple network model abnormal connection of each frequency band.The results demonstrated that in the 0.074-0.196 Hz frequency band,MDD patients had significantly reduced functional connectivity between CEN and SN and within the CEN,the abnormal connection in this frequency band could predict the HAMD score.The results showed that the abnormal connection of the triple networks in MDD was frequency specific.In the second part,the brain frequency-specific abnormalities of MDD patients were extended from the triple network model to the whole brain level to explore the topological abnormalities of the whole brain.It is not clear whether the abnormality of the restingstate functional network in MDD is frequency dependent,nor is it clear whether the topological attributes that characterize the structure of the brain network are frequency dependent.In this study,we used EMD method to construct the whole brain functional connectivity network under four frequency bands between MDD patients and health control,and used graph theory method to explore the abnormal topological properties of MDD patients brain network.The results showed that the MDD patients presented weaker small-world attributes in all frequency bands,the clustering coefficient is lower and the shortest path length is longer,and the local efficiency and global efficiency are reduced,the 0.021-0.042 Hz frequency band showed the most significant change.At the same time,the study on this frequency band found that MDD patients had abnormal frontal-temporal sub-network with reduced functional connectivity,and node degree centrality and betweenness centrality of some nodes are abnormal.The results showed that the abnormality of brain network topology in MDD patients was frequency specific.In summary,this study focuses on the functional connectivity of the triple network model and the macroscopic whole brain networks topological,these two studies revealed a frequency-specific abnormality pattern in the brain network of MDD,indicating the importance of frequency information in the study of rs-MRI in MDD.This study is an extension of previous studies on brain network in MDD,exploring brain network disorders with specific frequency,and providing a new perspective for studying the physiological and pathological mechanism of depression from the perspective of adaptive frequency division.