Complex Brain Network Analysis Of ADHD Based On Resting State FMRI

ADHD is a very common psychotic disorder in children. ADHD research has been a serious public health problem in our country and even in the world for more than ten years. The complex network analysis based on graph theory in recent years has greatly promoted the understanding of the network organization model. In this study, we created brain functional networks across each child based on RS-fMRI data of ninety-six children. Then analyzed the brain network anomalies caused by ADHD symptoms. Finally, we had carried on the preliminary exploration to predict the severity of the disease of ADHD based on the brain network parameters combined with pattern recognition algorithm.We have done the following research:First, we created brain functional networks across each child based on RS-fMRI data of ninety-six children (33ADHD patients and63typically developing controls). Then we got three nodal and six global network parameters of each child. The findings indicated that the brain network of children was small world network and was the same as adults. At the same time, we also got some "hub" brains which are critical for information transfer of brain networks and were similar to adults.Second, in order to investigate the basis of individual differences in the severity of inattention symptoms, we analyzed the correlation coefficients between each network parameter with IAS across the children based on the method of correlation analysis. The results indicated that significant positive correlations with IAS (P<0.01) were observed on the nodal parameters of seven regions, namely, bilateral caudate and thalamus, left middle cingulated cortex and lingual gyrus, and right hippocampus. In the further stepwise linear regression analysis, we observed that relatively satisfying prediction of the IASs could be obtained based only on the betweenness centrality of left MCC and the nodal efficiency of left caudate. These findings supported the importance of the seven regions for attention and demonstrated that local topological organization of the regions may substantially influence individual performances in attention-demanding situations.Finally, we predicted the individual level of inattention based on the complex brain network parameters and the method of support vector regression. In this study, we selected the features through the correlation coefficients between the network parameters and IAS. We constructed the IAS prediction model using linear kernel function and polynomial kernel function separately based on the method of support vector regression with leave-one-out cross-validation. The results showed that using fifty features of the strongest correlation with IAS and the polynomial kernel function could achieve the best effect of prediction. The correlation coefficient of the predicted IAS and the measured IAS was0.44.In summary, in this study we investigated the basis of complex brain networks in the severity of inattention symptoms based on RS-fMRI data and carried on the preliminary exploration to predict the severity of the disease of ADHD based on the brain network parameters. This study will not only promote us to understand the basis of the inattention, but also has the referential significance for the diagnosis of ADHD based on magnetic resonance imaging.