Identification And Effective Connectivity Temporal Lobe Epilepsy With Cognitive Impairment:Based On Granger Causality Analysis And Multivariate Pattern Analysis

Objective Temporal lobe epilepsy(TLE)is the most common form of refractory focal epilepsy,often accompanied by cognitive impairment(CI),such as memory,execution,and attention.At present,with the extensive application of resting state functional magnetic resonance imaging(rs-f MRI)technology,it has been found that patients with TLE with cognitive impairment are closely related to changes in brain structure and brain network,but these studies are mostly concentrated.In the case of undirected functional connections,regardless of the direction of influence,the direction of transmission of neural activity cannot be reflected.Effective Connectivity is based on the causal effect of one brain region on another.It can not only calculate the strength of the interaction between brain regions,but also obtain the direction of signal transmission between brain regions,which can be more accurately described.The interaction mechanism between different brain regions is of great significance for elucidating the neural mechanism related to the brain network of TLE with cognitive impairment(CI).This study uses brain network analysis methods such as granger causality analysis model,graph theory analysis,multimodal voxel analysis(MVPA).To investigate the mechanism of effective connection of cognition-related brain network in TLE-CI with patients.To conduct in-depth research on the effective connectivity brain function network mechanisms related to TLE with impairment.Methods This study included 32 TLE-CI with patients,30 TLE-NCI non-cognitive impairment patients,and 29 matched gender,age and education level Healthy control group.All subjects underwent resting state functional magnetic resonance scans and montreal cognitive assessment.In the first part of this paper,Granger causality model was used to analyze the effective connectivity modes within default network and core brain network of TLE-CI patients,TLE-NCI and healthy controls.In the second part of this paper,based on Granger causality model,the triple brain network(core brain network)of TLE-CI patients and healthy controls was analyzed respectively to explore the changes of the core cardio-brain network of TLE-CI patients;In the third part of this paper,the effective brain network topology attributes of TLE-CI and TLE-NCI patients were constructed based on Granger causality and graph theory analysis,and Multimodal voxel analysis was adopted.MVPA)were used to extract the topological attributes of TLE-CI and TLE-NCI effective brain network,and explore the value of the topological attributes of TLE-CI and TLE-NCI effective brain network in evaluating the classification and grading of TLE-CI and TLE-NCI diseases.Result 1.Changes in the effective connectivity within DMN bettween TLE-CI 、TLE-NCI patients and HCs group:(1)Compared with the HCs,patients with TLE-CI showed significantly enhanced effective connectivity from posterior cingutate cotex to medial prefrontal cortex and right parietal cortex to left parietal cortex.(2)Compared with the HCs,TLE-NCI within default network showed significantly increased effective connectivity right parietal cortex to left parietal cortex,however,significantly deceased effective connectivity right parietal cortex to posterior cingutate cotex.(3)Compared with the TLE-CI group,TLE-NCI group showed significantly enhanced effective connectivity from posterior cingutate cotex to medial prefrontal cortex.Moreover,TLE-NCI and TLE-CI patients had abnormal effective connectivity from RPC to LPC compared with healthy controls.2.Changes in the effective connectivity of the core networks in TLE-CI patients:(1)The effective functional connectivity from the posterior congulate cortex(PCC)seed point to other regions of the whole brain in the two groups showed that the different brain regions were mainly concentrated in the DMN The left fusiform gyrus and the right parahippocampal gyrus in the inner part;the effective connection from the PCC to the right dorsal prefrontal lobe(r DPFC)decreased in the TLE-CI group,suggesting the existence of effective functional connection from the DMN to the ECN Inhibitory connectivity;(2)The results of the effective connectivity from other regions of the whole brain to the seed point PCC in the two groups showed that the difference in the TLE-CI group was mainly in the left amygdala within the DMN;ECN in the TLE-CI group to the DMN in the DMN The right dorsal anterior cingulate with significant difference was extracted,that is,the effective connection of the right dorsal anterior cingulate to the PCC was decreased,suggesting that the ECN-DMN connection was shown as an excitatory connection in the TLE-CI group;(3)the two groups Results of active connectivity from other regions of the whole brain to the ECN showed that in the TLE-CI group,abnormalities in the effective connectivity of the left superior parietal lobule within the ECN were observed;the effective functional connectivity of the right precuneus to the right dorsal prefrontal cortex was enhanced,suggesting that ECN-DMN connections are excitatory connections in TLE-CI;there was no significant differences were observed in the connections from other brain regions to r DPFL between the two groups;(4)the right dorsal forebrain between the two groups was not significantly different The results of effective functional connectivity from the insula seed point to other regions of the whole brain showed that the abnormal effective connectivity from SN to DMN was observed in the right middle temporal gyrus to the right dorsal anterior insula.The SN to the DMN showed excitatory connectivity;no significant differences were observed between the two groups comparing effective connectivity to the SN from other regions of the whole brain.3.Research on the extraction and classification of topological attributes of effectively connected brain network :(1)abnormal effective topological attributes can identify the brain networks of tl-nci patients and tl-ci patients and healthy controls with the accuracy of 86.82% and 83.05%,respectively.SVM classifier has obtained high accuracy and sensitivity.The classification accuracy of TLE-CI and TLE-NCI was 75.81%,which might be related to the brain lesions of TLE subjects with cognitive impairment.Support vector machine is used to classify three groups of subjects by classifier,and the result of classification is better.(2)There were no significant differences in t LE-CI,TLE-NCI and HCs among the three groups(P >0.05).Among the local features,significant differences(P <0.05)were found in the middle frontal gyrus,fusiform gyrus,supoccipital gyrus,parahippocamal gyrus,thalamus,putamen,posterior cingulate gyrus,lingual gyrus,precuneus,parietal gyrus,superior marginal gyrus and insula.Table 10 shows the information of all brain regions with significant differences.It can be seen that the parahippocampal gyrus of the three groups showed significant differences in local efficiency,degree and node intensity,and the insula showed significant differences in such as Page Rank centrality and entry degree.The superior occipital gyrus showed significant differences in the degree of emergence and Page Rank centrality,and the other three groups also showed significant differences in the anterior cingulate gyrus,posterior cingulate gyrus,thalamus,lingual gyrus,precuneus,fusiform gyrus,superior temporal gyrus,etc.The differences were mainly found in default network,salience network,central executive network and visual network.Conclusion 1.The effective connectivity pattern within DMN is different between TLE-CI patients and TLE-NCI patients,suggesting that the pathophysiological mechanisms may be different between patients.Increased effective PCC to MPFC and RPC to LPC connections within DMN in TLE-CI patients may be an ineffective compensation.To provide new insights into the pathophysiological mechanism of DMN in TLE-CI and TLE-NCI patients.2.In the TLE-CI patients group,there were abnormal effective connections in the core brain network,and excitatory effective connections in ECN-DMN and SN-DMN,suggesting that the increased information outflow of ECN-DMN and SN-DMN might be a compensatory effective functional connectivity.3.Most of the significant abnormalities were concentrated in four areas,such as degree,node intensity and local efficiency.The most influential network topology attribute was degree,and the variability of effective brain network topology attribute may be an important indicator for studying TLE-CI and TLE-NCI patients.SVM classifier can help to distinguish TLE-CI and TLE-NCI population from healthy control population,thus providing help for TLTLE-CI and TLE-NCI auxiliary diagnosis.