The Applied Research Of Asymmetric Functional Connectivity Neteorkd In AD Early Diagnosis

Alzheimer’s disease(AD)also called Primary Alzheimer’s disease,is serious with advanced cognitive and memory dysfunction for clinical performance.Mild cognitive disorder(MCI)is an intermediate state between AD and heathy people.These diseases have a long incubation so it is hard to find.Moreover,the diseases are serious and the hurt is irreversible.China has a large population and has an ageing social risks,so it is a hard work to prevent Alzheimer.Therefore,to find the early features of the diseases so that we can treat them in time and save costs is the emphasis of researches.Now,AD aided diagnosis based on magnetic resonance has achieved some appreciable results.However,most of the researches were based on the structure image,the brain structure changes means it is too late to treat it in time.If we can find the cognitive dysfunction as the diagnostic basis,we will treat the disease in the early time.So it is important for the early aided diagnosis to find the anomaly index of the cognitive function.Hemispheric asymmetry is the differences between left hemisphere and right hemisphere both in anatomical structure and in cognitive function.In this study,we used graph theory to explore the hemispheric asymmetry in AD so that it can served the AD aided diagnosis and improved the classification accuracy.The main work of this study runned with the following steps:(1)Deffer from the traditional methods,we made the symmetric brain template to construct hemisphere brain networks,calculated functional connection strength and network properties,and then calculated the laterality index of function connection strength and network properties.(2)By using statistical analysis,we screened the features to AD aided diagnosis.Moreover,we explore the physiological signification of the features.(3)According to the features,we made the feature space,trained the classification model with SVM classifier and test the classification model with leave one out cross-validation.(4)We used ADNI database to verify the idea we proposed.The result showed that the hemispheric asymmetry of AD patients was different from the healthy people.The features were similar with some previous researches.Importantly,we improved the classification accuracy to 85.71%,sensitivity to 87.06% and specificity to 84.34%.In this study,we compared the accuracy of using functional connection strength,network properties and their laterality index to classification.In conclusion,the laterality index was helpful for classification,especially for the classified between MCI and heathy people.Therefore,lateralization factors were meaningful for AD early diagnosis.