Human Brain Functional Connectivity Prediction Based On Complex Network Measures

The complex relationship between human brain functional networks and structural networks remains a central focus of the emerging field of neuroscience.Although mounting evidence has demonstrated statistical correlation between functional connectivity(FC)and structural connectivity(SC),how SC shapes or constrains FC is still challengeable and thus has motivated a number of computational models to investigate the function-structure relationship.This thesis mainly focuses on communication measures derived from SC to explore the influences on FC.The main work of this thesis can be described by the following set of points:1.Several network measures extracted from SC are studied.Besides,a new simple measure——Accessibility——is extrated from SC based on the two present communication measures——Search Information and Path Transitivity;2.A linear model based on Accessibility measure is proposed to predict FC.Compared with other linear models on the basis of the other network measures and a neural mass model,our results examined on three different-resolution datasets indicate that the predictive capacity of the linear model based on Accessibility measure is superior to others;3.A multi-measure nonlinear model which consists of many network measures is developed.The proposed nonlinear model is tested on three datasets.Experimental results show that the multi-measure nonlinear predictor performs better than any other single-measure nonlinear model and the neural mass model.