Identifying The Alteration Of High-level Visual Cortical Network In Individuals With Major Depressive Disorder:A Study Based On Functional Connectivity Of Magnetic Resonance Imaging

Part I Study on the Differences and Dynamic Changes of Visual Cortex Network in Individuals with Major Depressive Disorder Based on Functional Magnetic Resonance ObjectiveThe visual cortex is the initial stage of brain information reception and processing.The purpose of this part is to use the resting state functional magnetic resonance technique combined with advanced visual network map to study the changes of visual network characteristics in patients with major depressive disorder,and to solve the problem of comprehensive analysis of visual networks in previous studies.Materials and methods1.Research object:According to the Diagnostic and Statistical Manual of Mental Disorders.4th Edition(DSM-?)Diagnostic criteria for depressive episodes and 24 Hamilton Depression Scale scores,86 patients with depression(27 males,59 females,median age:30 years)were collected as the disease group.At the same time,73 healthy volunteers(17 males,56 females,median age:29 years old)matched with age,sex and education level of the disease group were selected as the control group.All subjects in the two groups had no history of other mental disorders,history of drug dependence,history of brain organic disease,cognitive impairment,severe heart,liver,kidney,pulmonary dysfunction,and magnetic resonance scan contraindications,and both were right.Good hand.2.Data acquisition:Data was acquired using Philips 3.0T superconducting magnetic resonance 8-channel head-moving coil.First,the 3D-T1WI images were obtained by using the turbo field echo(TFE)sequence.The scanning parameters as follows:TR = 8.2 ms,TE TE 3.8 ms,flip angle = 7°,bandwidth = 191 Hz,FOV = 256 x 256 mm2,voxel = 1.0 × 1.0 x 1.0 mm3.The resting-state fMRI(rs-fMRI)scan uses a field-echo echo-planar imaging(FE-EPI)sequence with the following scanning parameters:TR = 2000 ms,TE = 30ms,flip angle = 900,bandwidth = 4131 Hz,FOV=240 × 240 mm2,voxel = 3.4× 3.4×3.4 mm3.Matrix = 64 x 64,Slice number is 33 with layer-by-layer,Slice thickness = 4 mm,Slice gap = 0.6 mm,NEX=1,time points = 240,scanning time = 8 minutes.Collect general information(including gender,age,education level,treatment history,medication history,etc.),and Hamilton Depression Scale(HAMD)score,Hamilton Anxiety Scale(HAMA).3.Data preprocessing:Rs-fMRI data was preprocessed using SPM12(http://www.fil.ion.ucl.ac.uk/spmy),The preprocessing steps of data include rejecting the first 10 time points,time layer correction,and head movement Correction,image registration,spatial smoothing,voxel-wise detrending,filtering,and regression out covariates.4.Visual network cortical localization and network construction:The ventral and dorsal visual network cortical localization is determined based on the previously published brain coordinates.For the ventral and dorsal visual networks,we adopted previously established statistical maps of ventral and dorsal pathways based on ALE analysis.In the generated ALE cluster,each peak coordinate was used as a central node to create a spherical region of interest(ROI)with a radius of 5 mm.The nodes of the ventral and dorsal flow networks are defined according to these advanced visual brain ROls.In addition,the Talairach spatial coordinates of the ROI of the previously published bilateral primary visual cortex were converted into MNI spatial coordinates to construct a complete ventral and dorsal visual network node.The resting-state functional connective(RSFC)between the two ROIs were obtained by calculating the rs-fMRI time series correlation of each of the two ROIs as the edge of the network.Construct a complete ventral and dorsal visual network.5.Visual network parameter calculation:The static network measurements were established based on the RSFC of full-length rs-fMRI time series using previously described network quantification method.First,the RSFC matrices were binarized with a spectrum of network densities(5%-40%;interval,1%)to ensure that the networks were comparable among different individuals.Then,for each matrix across the density range,the three network parameters of characteristic path length,clustering coefficient and small-worldness were calculated respectively to assess the abilities of integrated and segregated information processing of the visual networks.For a better evaluation of the network topology,a normalization was done for the three parameters by comparing to 1000 randomly generated networks,which respectively generated the?,?,and?.Moreover,the mean connectivity strength was calculated by accumulating the RSFCs for each node,then averaging across all the nodes within a network.6.Statistical analysis:The demographic and clinical information of the subjects were evaluated using a statistical test of the Statistical Software and Service Solutions 23.0(SPSS 23.0).The Mann-Whitney u-test was performed for the comparison of age,education level,HAMD and HAMA scores between the MDD and the control group.Gender differences between groups were assessed using the X2 test.A statistical significance level of p<0.05 was adopted.For the between-group comparisons of static properties of dorsal and ventral visual networks,two sample t-test was performed for the density-wise comparisons of network properties across 5%-40%,and a p<0.01 with false discovery rate correction was used to correct for the 36 times of multiple comparisons.The area under the curve(AUC)value was additionally extracted over the whole density range for each individual and compared between two groups using the two sample t-test(or Mann-Whitney u-test for non-normally distributed data).For the comparisons of mean connectivity strength,dynamic variability,two sample t-test(or Mann-Whitney u-test for non-normally distributed data)was used.Here,for the network measurements of the same nature,a p<0.05 with Bonferroni correction was adopted to correct for the number of tests(e.g.,for static connectivity measurements,p<0.05/4 static network metrics/2 visual networks=0.00625).In addition,all the above comparisons were adjusted for age,gender,education level,and head motion(i.e.,FWD)to control their confounding effects.ResultsAge(p?0.101),gender(p=0.256),education level(p=0.224),and head motion during MR scan(p=0.225)were matched between MDD and control groups.Compared with the control group,the clustering coefficients of the two visual networks in the MDD group were significantly increased(dorsal:p=0.002;ventral:p = 0.004),and small-worldness were also significantly increased(dorsal:p = 0.001;ventral:p = 0.002).Compared with healthy controls,both the dorsal and ventral visual networks were found significantly elevated mean temporal variability of FC in MDD group(p<0.001 and P=0.001,respectively).Moreover,the mean connectivity strength of the two visual network was significantly lower in MDD group compared with control group(p<0.001).Conclusion1.For the static connectivity analyses,a significantly increased clustering coefficient,small-worldness along with lower network connectivity strength were observed in the dorsal and ventral visual networks in MDD patients,indicating that the internal structure of the visual network of MDD patients is more optimized and economical.Local information processing is more efficient,and network integration capabilities were well preserved.2.For the dynamic connectivity analyses,results revealed significantly increased temporal variabilities of both the ventral and dorsal networks in MDD patients.This result might indicate that visual network in the MDD group with higher level of functional activity.Part II Study on the Abnormality of Visual-Attention Network and Its Correlation with Disease Severity in Patients with Major depressive disorder Objective Exploring the changes of attention network in the regulation of visual network in patients with major depressive disorder,and further study the relationship between visual network internal and visual-attention network changes and the severity of depression disease,and then explore the possible mechanism of visual network changes in the development of depression disease,and provide objective imaging basis for clinical diagnosis and treatment.Materials and methods1.Research object:The same to part ?.2.Data acquisition:Data was acquired using Philips 3.0T superconducting magnetic resonance 8-channel head-moving coil.The scanning sequence and parameters are the same to part ?.3.Data preprocessing:The same to part ?.4.Visual-attention network cortical localization and network construction:For the dorsal attention network,the coordinates corresponding to DAN were defined based on the group-averaged cortical network seeds from 1000 brains study,Similarly,the spherical ROIs of DAN were created by centering at each coordinate with a radius of 5 mm.There are a total of 12 ROIs,which are regarded as nodes.The paired static state functional connections between the DAN and dorsal networks and the ROIs of every two networks between the DAN and ventral networks are regarded as edges,thus constructing corresponding The internet.Then,a functional connection matrix between DAN-dorsal(31×12)and DAN-ventral(22×12)networks is constructed for each subject.5.Connection calculation between DAN and ventral/dorsal network:The inter-network connectivities between DAN and ventral visual network as well as between DAN and dorsal visual network were measured using the multivariate distance correlation.In the current study,we used a modified distance-dependent statistic to calculate the connection,which was considered to be unbiased by the number of ROIs in each network.6.Statistical analysis:Statistical tests were performed using a statistical software package(Statistical Product and Service Solutions 23.0,SPSS 23.0).For the comparisons of the inter-network connectivity(i.e.,distance correlation between DAN and the two visual networks),two sample t-test(or Mann-Whitney u-test for non-normally distributed data)was used.Here,for the network measurements of the same nature,a p<0.05 with Bonferroni correction was adopted to correct for the number of tests.In addition,comparisons were adjusted for age,gender,education level,and head motion(i.e.,FWD)to control their confounding effects.To evaluate the clinical significance of the network analyses,we assessed the correlation in MDD group between HAMD score and above network measurements using parital correlation with age,gender,education,head motion,and history of medication treatment added as covariates.Similarly,a p<0.05 with Bonferroni correction was used to correct for the number of statistical tests for the network measurements of the same nature.To further clarify the possible confounding effect of several nuisance factors(e.g.,anxiety)on the positive network findings,we additionally performed a regression analysis which included HAMA,age,gender,education,head motion,and history of medication treatment as independent factors.Here,p<0.05 was used as an exclusive significance level.ResultsWe found that MDD patients showed significantly lower distance correlations of DAN seeds to both dorsal and ventral visual network ROIs(p=0.004 and 0.013,respectively).Moreover,the correlation analysis between HAMD score and network parameters showed that the DAN-dorsal visual network connection was negatively correlated with the HAMD score(r =-0.298,p=0.007).Regression analysis showed that the HAMA score had a significant contribution to the change in dorsal-DAN network connectivity(p=0.35).Conclusions1.Through the comparative analysis of the connection between DAN and dorsal/ventral visual network of normal people and MDD patients,it was found that the visual network of depression patients receives less regulation from the attention network.It may mean that the visual network of MDD patients receives less supervision from the advanced center and adapts to a more autonomous state in inter visual network.2.The results of the correlation analysis indicate a negative correlation between the disease severity and the dorsal-DAN internetwork connectivity decrease.In addition,regression analysis showed that the contribution of HAMA to the dorsal-DAN internetwork connectivity change cannot be excluded,further suggesting that the comorbidity between depression and anxiety was also related to dorsal-DAN internetwork dysconnectivity.