Network analysis on neuro-imaging data

Abstract

Neuroimaging is an effective technique to examine the structure and connectivity of human brains. Also, neuroimaging data has been widely used in clinical diagnosis and research areas. This project is to develop a method to predict the age group that the brain belongs to using neuroimaging data. Neuroimaging data were collected from public sources and processed using existing pipelines. Default mode brain network was constructed from processed neuroimaging data. Network analytics method was applied and several network features such as efficiency, clustering coefficient were selected and calculated. The calculated data was used for classifier training purposes. Three different multiclass classifiers, OneVsOne, OneVsRest, and K-NN classifiers were trained. Evaluation of performance was done on each trained classifier based on calculated accuracy and F1 score. This was to compare and find out the most suitable classification algorithm for brain age group classification and prediction. From the result obtained, the classifiers can have high accuracy and make accurate classification and prediction of the brain age group.