The Study Of Structural Connectome For Brain Aging

The aging of global population is becoming a serious problem. As one of the world’s most populous country with aging population, China has entered the stage of rapidly ageing, the individual aging process inevitably occur in the brain structure, functional morphology, which is often called brain aging. At present, many studies have pointed out that the normal brain aging and various neurodegenerative diseases have a common brain degeneration model, so it is important to study brain aging for the early prevention, diagnosis and treatment of disease,which is related to brain aging.In this paper, we constructed brain cortical thickness network and white matter fiber network based on multimodal magnetic resonance imaging(T1WI, DTI) data in normal aging respectively. Then, we used graph theory method to analysis network topology parameters within the aging process, and selected morphological structures and network topological parameters which are highly correlated with subject’s real age. We established a prediction model of brain aging,the main contents of the research include the following aspects:(1) Based on T1 WI imaging data of 75 normal aging subjects, we first calculated the cortical thickness of each subject, then the data is divided into four groups according to their age with 5 years interval. Each group is based on Pearson correlation coefficient of different brain regions to construct the cortical thickness network, and graph theory method was applied to analysis network topology parameters. Finally, based on the cortical thickness topological parameters,we made a series of statistical analysis. The result showed that the changes of brain network topological properties is nonlinear in the aging brain. With the increase of age, local information processing ability increases after the first decrease, while global information processing ability decreases after the first increase.(2) Based on DTI image data of same cohort, we first extracted each subjects white matter tracts, after combined with AAL template, we achieved white matter network.Then, we used graph theory method to calculate the local and global network topological parameters. Finally, based on the cortical thickness and network topological parameters, we performed a series of statistical analysis. The result showed that edge density of white matter fiber network decreased with the increase of age. Although the brain backbone networks are in continuous reorganization, the core backbone network maintain a relatively stable state.(3)Based on the above results, we obtained morphological structures and network topological parameters. We used feature selection and principal component analysis(PCA) for dimensionality reduction. After that, multivariate linear regression and BP neural network methods were used to establish prediction model to predict brain age. Because of limited dataset, we used the leave one out cross validation method to test the prediction accuracy. The result showed that the performance of our approach is better than several published methods.