The Role Of Brain Network Analysis Based On Graph Theory In The Evaluation Of Functional Impairment Of Brain Network In Tremor-dominant Parkinson's Disease

Objective:In this study,a whole-brain functional network was constructed on the basis of the resting state-functional magnetic resonance imaging(rs-fMRI)technology with the network analysis method of graph theory.The cerebral functional network status of the tremor-dominant Parkinson's disease(TP)group and the healthy controls(HC)group were observed,and the differences between the two groups were compared.The change characteristics of brain functional network in the TP group were analyzed.To explore the role of brain functional network constructed based on rs-fMRI technology and network analysis method of graph theory in the evaluation of functional impairment of clinical TP brain network.Materials and methods:32 TP patients and 31 age-,gender-and education-matched healthy volunteers were recruited and underwent rs-fMRI examination by 3.0T MR scanner.The brain network analysis software based on graph theory GRETNA(http://www.nitrc.org/projects/gretna/)was used to preprocess the fMRI data,construct brain network,analysis network properties and analysis statistical differences of the two groups one by one.The following preprocessing steps were taken,including removing the data of the first 10 time points,slice timing corrections,head motion corrections,realignment,normalization,segmentation,detrending,filtering and spatial smoothing.After head motion corrections,the data of three TP cases and one HC case were excluded because of the excessive head motion(maximal rotation>3° and/or maximal translation>3mm in x,y,or z).Finally a total of 29 cases in TP group and 30 cases in HC group were enrolled in this study.After that,the whole brain functional network was constructed for the two groups of subjects respectively with the template of anatomical automatic labeling(AAL)116.For network analysis,we didn't use a single sparsity(threshold)method,but a sparsity range of 0.05-0.5,with an interval of 0.02.In the range of the sparsity,we calculated the global property indexes[including:clustering coefficient(Cp),characteristic path length(Lp),small-worldness(Sigma),global efficiency(Eg)and local efficiency(Eioc)]and the node property indexes[including:nodal clustering coefficient(Ci),nodal efficiency(Ei),nodal betweenness centrality(BCi)and nodal degree centrality(DCi)]of the two groups respectively.In the network statistics comparison section,for global property indexes,we calculated its Area Under Curve(AUC)value within the range of values of sparsity 0.05?0.5.Relative to the selection of single sparsity(threshold),AUC provided a comprehensive evaluation for the property.The AUC value of the global property index was compared by the inter-group two-sample t test.When P<0.05,the difference between the two groups was statistically significant.As for the node property indexes,the inter-group two-sample t test was used for comparison.False positives were corrected by false discovery rate(FDR)because there were multiple brain regions between the groups to compare the node property indexes and false positives were increased.When P<0.05,the difference was statistically significant.Results:The comparisons of the global property indexes between TP and HC groups:in the range of 0.05?0.5 sparsity,both TP and HC groups exhibited typical features of small-worldness,and the clustering coefficient(Cp),local efficiency(Eloc)and small-worldness(Sigma)of TP group were significantly lower than HC group through AUC analyses.When P<0.05,Cp,Eloc and Sigma were statistically significant between TP and HC groups.The comparisons of the node property indexes between TP and HC groups:1.nodal clustering coefficient(Ci):Compared with HC group,the decreased brain areas in TP group were widely distributed in cerebellum,frontal lobe,insula,occipital lobe,parietal lobe and temporal lobe,while no increased brain areas were found in TP group compared with HC group.2.nodal efficiency(Ei):Compared with HC group,the decreased brain areas in TP group were distributed in occipital lobe and parietal lobe,while the increased areas in TP group were distributed in cerebellum,pallidum,frontal lobe,temporal lobe and parietal lobe.3.nodal betweenness centrality(BCi):Compared with HC group,the decreased brain areas in TP group were distributed in occipital lobe,parietal lobe and temporal lobe,while the increased areas in TP group were distributed in thalamus,frontal lobe,insula and parietal lobe.4.nodal degree centrality(DCi):Compared with HC group,the decreased brain areas in TP group were distributed in occipital lobe and parietal lobe,while the increased areas in TP group were distributed in cerebellum,frontal lobe,temporal lobe and parietal lobe.The results showed that there were differences between the TP group and the HC group in the node property indexes of the above brain regions,and the P values were all<0.05,indicating a statistically significant difference.Conclusion:1 Compared with HC group,the small-wordness property(Sigma),clustering coefficient(Cp)and local efficiency(Eloc)of TP group were significantly reduced,and the differences between the two groups were statistically significant.This indicated that the functional differentiation degree local information transmission ability of brain network in patients with Parkinson's disease with tremor were lower than that in healthy control group,suggesting that some functions of brain network in patients with Parkinson's disease with tremor are impaired.2 Compared with HC group,nodal efficiency(Ei)of right globus pallidus,nodal efficiency(Ei)and nodal degree centrality(DCi)of bilateral cerebellum and nodal betweenness centrality(BCi)of right thalamus in TP group were significantly higher than those in HC group,and the differences between the two groups were statistically significant.Combined with the literature,it is speculated that the increased node properties of these brain areas may be related to the occurrence of tremor in patients with Parkinson's disease.3 The results of this study showed that the brain function network of TP group had certain characteristics compared with HC group,indicating that the brain function network constructed by rs-fMRI technology and network analysis method of graph theory had potential application value in the evaluation of brain network function impairment in patients with tremor-dominant Parkinson's disease.