The Dynamic Analysis System Of Brain Function Connection Based On FMRI

Nowadays,more and more researchers have used computational methods to study brain function.Brain science has become one of the fastest growing disciplines.The study of brain functional connection is an extremely hot topic in the field of brain science,which helps people to understand human brain function and working mechanism in depth,to further understand and develop the human brain,and to help diagnose and treat various brain diseases,Therefore,it is of great importance to neuroscience and clinical medicine.Based on the dynamic analysis of the functional connection of the brain,this paper proposes the research framework of the dynamic analysis of the functional connection of the brain,and uses the related methods of machine learning to study the dynamic connection of the functional connection of the brain.The main work of this paper is as follows:1)We compare and evaluate the effects of time series segmentation method,Bayesian connectivity change point model(BCCPM),the method based on Fisher linear discriminant Criterion,k-means and spectral clustering method,on brain functional MRI data.The experimental results show that BCCPM and the method based on Fisher linear discriminant criterion have better effect on fMRI datasets than others and can won’t detect too many transition points.2)A novel local feature extraction method LBEM based on binary coding was designed.On a real dataset of attention-deficit/hyperactivity disorder(ADHD)children and normal control children,we compared the framework which lacks LBEM with our research framework.The experimental results show that LBEM can effectively extract the identifiable local features of the region of interest(ROI)in the brain.3)The paper introduced the extreme learning machine(ELM)algorithm and Kernel ELM algorithm.In the same research framework,the classifier uses ELM and KELM respectively to perform classification experiments on ADHD datasets.The results show that the classification accuracy and stability of the KELM classifier are better than that of the ELM classifier,and the speed is slower than the ELM.4)Design and implement a brain functional connection pattern recognition system intergrating image preprocessing,dynamic detection,feature extraction and classifier.