| Populations around the world are rapidly aging. In developing countries,population aging is progressing faster than developed countries. Aging is one of themost important known risk factor for dementia. After age65, the likelihood ofdeveloping dementia roughly doubles every five years. Dementia is an irreversibleneurodegenerative progress. It will cause memory loss and other cognitive deficitsthat are sufficiently severe to interfere with the person’s ability to perform ordinaryactivities of daily living. Currently, there are no treatments available that can cure, oreven alter the progressive course of dementia, although numerous new therapies arebeing investigated in various stages of clinical trials. Therefore, secure and healthyaging has been widely seen as one of the most important event in the society. Earlydetection and diagnosis of the cognitive status of old people has been an inevitablehealth issues. In recent years, non-invasive neuroimaging techniques have provided anew perspective on structural and functional connectivity patterns of the human brain.Diffusion Tensor Imaging (DTI), a non-invasive technique that can investigate whitematter microstructure, has gained acceptance in medical imaging. It enables recoveryof the structure of white matter within the human brain by tractography. At the sametime, it intuitively reflects the changes of white matter connectivity networkassociated with decline in cognitive test score.In this thesis, on the basis of DTI, we analyzed the properties of structural brainnetwork by selecting the characteristics of structural network which were highlycorrelated with cognitive performance to estimate the prediction models. Based on theestimated models, the score of cognitive performance of subjects could be evaluatedobjectively according to their DTI. The main contents of the research include thefollowing aspects:(1) We used DTI of52healthy elders to construct brain structural network. Inorder to obtain the characteristics of structural brain network, we analyzed connectionmatrices using graph theory and diffusion tractography. We selected significantfeatures based on the correlation analysis between the characteristics of brain networkand subject’s Mini-Mental State Examination (MMSE) score. These features wouldthen be used to estimate the model based on machine learning algorithm to predict thecognitive performance. Finally, the performance of prediction model would be analyzed and discussed.(2) We employed DTI of39aMCI subjects to construct brain structural network.In order to obtain the characteristics of structural brain network, we analyzedconnection matrices using graph theory and diffusion tractography. We selectedsignificant features based on the correlation analysis between the characteristics ofbrain network and subject’s MMSE score. These features would then be used toestimate the model based on machine learning algorithm to predict the cognitiveperformance. Finally, the performance of prediction model would be analyzed anddiscussed.(3) Discuss the classification efficiency of the prediction model that combined thecoginitive performance predictive model based on normal aging dataset with thecoginitive performance predictive model based on aMCI dataset. A “leave-one-outcross validation” method was used to estimate the classification performance of thetwo prediction models to classify normal aging and aMCI.In conclusion, the academic value of this thesis is mainly reflected in structuralnetwork metrics derived from DTI can be as new biological marker pointers toestablish prediction models for cognitive performance in health aging and aMCI. Andprediction models can also achieve the classification of normal aging subjects andaMCI patients. Futhermore, these structural network features can provide relevantinformation for clinicians, can explain which brain regions changes caused by agingand disease and also can offer the change information between corresponding brainregions. |
PDF Full Text Request