Changes Of The Brain In Alzheimer's Disease And Mild Cognitive Impairment Base On DTI:A Brain White Matter Fiber Network Study

Alzheimer's disease(AD) is the most common type of dementia. The major reason why AD patients can't get effective treatment so far is that the pathophysiological mechanism of AD is still unclear. The patients show memory deficits and communication disorder at the early stages of the illness, and then lose activity ability and memory with the development of the disease, which cause heavy economic burden to their family and the society.Mild cognitive impairment(MCI) is known as the intermediate state between dementia and normal state, which always is accompanied by memory problems or other cognitive degeneration. Since there being higher risk for MCI patients meeting disease transformation to AD than normal person, the researchers have paid more attention to the pathological mechanism of the transformation in recent years. The exact development mechanism for the disease is not clear, although many important results have been obtained. Further study is needed.With the development of magnetic resonance image(MRI), the changed brain functional or white matter fiber regions could be found, which takes new research direction for the occurrence and development of AD. Most researches about the changing of white matter fibers focused on the regions of interesting(ROI) or the specific network models such as small world networks or free-scale networks. Since there were many results, the inconsistent conclusions were always got. We found loosing connection of the hubs, the central node of a network, is almost consistent with the phenomenon after a large number of literature research. So, we propose the hypothesis that the rich-club network model, which is a hierarchical network base on the hubs, could be better used to describe the changes of patient brain. The main research results and innovations are summarized into the following three parts:The first part evaluated the changes of the hiera rchical network for AD and MCI the patient. Based on the image registration and segmentation of the structural MRI and diffusion tensor image(DTI), we parted the cerebral cortex into 90 regions. With the help of determined fiber tracking, we get networks which describe the fiber number between the regions. By calculating the networks parameters suc h as rich-club coefficient and connection strength at different brackets, we found that:(i) the loosing connections of AD occurred at the second and the third structure;(ii) MCI occurred just at the third structure. That may be the reason why AD is more serious than MCI.The second part introduced the biomarkers which can distinguish from normal controls(NC) with high accuracy. Based on the conclusions we get in the first part, we suppose the biomarkers could be found to describe AD happens. By calculat ing the betweenness centrality of each rich node, which describe the central role in the information transmission, and connection strength of the node connected to the rich node. We find that the biomarkers, the betweenness centrality of left putamen(PUT) and connection strength between left triangular part, could distinguish AD with an accuracy of 89.09%(the mean sensitivity from the classification by support vector machines(SVM)).The third part investigated a method of data mining which used for improving the signal to noise ratio in functional MRI networks, including basic assumptions of the calculation model, data processing and verification of the model. Since the results were not satisfied, it was reasonable to solve the problem. Further work to be performed.