| ObjectiveAutistic spectrum disorder(ASD) has been defined as a neurodevelopmental disorder with associated deficits in executive function, language, emotional, and social function. Children with ASD have severe impairments in communication and social interaction, causing great burden and economic load to their family. Early diagnosis and early intervention for children with ASD may ameliorate symptoms and improve the quality of life. But these must be done in the early childhood of ASDs. In this thesis, brain effective connectivity networks were constructed to study the neural changes in children with ASD, and these studies may provide basis for clinical diagnoses.Methods1 Data: The 19-channel EEG data were recorded from children with ASD(N = 17, aged 30± 0.9 months) and healthy control group(N = 17, aged 29± 1.2 months) during an eyes-open resting condition.2 The EEGs causal effective connectivity was characterized by Granger Causality. Directed transform function(DTF) was applied in the calculation of connectivity matrix. In this research, EEGs were analyzed in 5 frequency bands: theta band(4-8Hz), alpha band(8-13Hz), lower-beta band(13-24Hz), higher-beta band(24-30Hz) and gamma band(30-60Hz).3 Brain networks of the two groups were constructed in the range of the thresholds to the connectivity matrixes, T: 0.02 ~ 0.06, step: 0.001.4 The global network characteristics were compared between two study groups using graph theoretical analysis methods, including the average degree of the network, the global efficiency and the average local efficiency.5 The local efficiency difference and its spatial distribution between two groups' brain networks were analyzed under the optimal threshold.6 Network nodes in gamma band were sorted according to the nodes' centrality(firstly degree centrality and then BC centrality) in combination with the K-core decomposition method.7 Statistical analyses: the statistical data was analyzed with independent samples T-test.Results1 The whole brain DTF sum values showed significant difference between two groups only in gamma band networks(p<0.05).2 Compared with the brain networks of the control group, the ASD group showed lower global and local efficiency and lower node degree. Within a certain range of threshold value, there were significant differences.3 The optimal network thresholds for 5 bands were: Tgamma = 0.037 for the gamma band, ThighBeta= 0.036 for the higher-beta band, TlowBeta= 0.042 for the lower-beta band, Talpha = 0.036 for the alpha band and Ttheta = 0.042 for the theta band. The local efficiency of control group gamma band networks was significantly higher than that in ASD group(p<0.01). In the parietal region, the gamma, higher-beta, lower-beta and alpha band networks' local efficiency showed significant group differences(p<0.05).4 The important nodes in the gamma band networks were sorted based on the nodes' centrality combined with k-core decomposition. The important nodes of ASD networks were O1, O2, PZ, P3 and P4, concentrated in the parietal occipital region; while the control network important nodes distributed in occipital region and prefrontal region for O1, O2, FP2, FP1 and T6.Conclusions(1) Analysis of network properties revealed differences between ASDs and healthy controls and these differences were more significant in higher frequency bands(particularly in the gamma band).(2) Compared to the healthy controls, ASDs had lower brain network density. A reduction in both global and local efficiency of ASD networks may reflect impaired brain connectivity in children with ASD.(3) The local efficiency of the networks in two groups showed most obvious difference in the parietal-occipital region. The gamma band brain network showed transferred core nodes in the ASD group: the frontal nodes were more important in the network of normal children than that those in children with ASD, which may be related to the impaired prefrontal area in children with ASD. |
PDF Full Text Request