Study On Predictive Methods Of Brain Functional Connectivity

Brain functional connectivity prediction is one of the important research directions in the field of neuroscience,which has significant meanings in both theoretical research and clinical application.At present,the research on brain functional connectivity prediction mainly focuses on static brain network,ignoring the temporal evolution information of brain network,and most of the research does not consider the spatial correlation of voxel level in brain region,so there are certain deficiencies.In order to predict brain function connectivity with disease progression and normal aging,this study focuses on the prediction of brain functional connectivity from a time and space perspective.Thereby it can achieve early intervention in mental illness and help doctors make clinical decisions.The study mainly contains the following parts:Aiming at the characteristics of dynamic changes of brain functional connectivity,a brain functional connectivity prediction method based on dynamic bayesian hierarchical model was proposed.This method overcomes the limitations of static methods and captures the time-varying characteristics of the brain network.At the same time,it can be combined with patient-related factors,such as disease status,gender age,etc.,to provide accurate predictions of the patient's future brain functional connectivity.Aiming at the complex spatio-temporal correlation of longitudinal neuroimaging data,a prediction method based on combined model was proposed.It overcomes the shortcomings of existing methods that only consider time correlation,predict a single time period,and avoid voxel-level connections.At the same time,the spatial information between adjacent voxels is used to further improve the prediction accuracy.It is also suitable for predicting the functional connectivity with the normal aging of individuals,and deepening the understanding of the human brain.In this research,the proposed algorithm is verified and analyzed on the ADNI dataset.The results of this study indicate that the performance of the proposed algorithm is better than the traditional prediction method,Therefore,the effectiveness of the algorithm is verified.