Analysis And Classification Of Changes In Brain Functional Connections In Patients With Alzheimer’s Disease

Alzheimer’s disease is a neurodegenerative disease,the main symptom is a decline in emotional,language,and cognitive functions.With the aging of the population in recent years,the probability of its onset has gradually increased.For the diagnosis of Alzheimer’s disease,the traditional diagnostic methods are mainly psychological tests and clinical inquiry diagnosis.This method has a high rate of false positives and is easily misdiagnosed,thereby delaying the optimal period of disease control.Therefore,earlier diagnosis and prediction of Alzheimer’s disease is very important for the control and treatment of the disease.In recent years,with the development of science and technology,functional magnetic resonance imaging technology has been widely used as a new technology.This method can reflect changes in nerve activity by detecting changes in blood oxygen level-dependent signals at rest.By calculating the time correlation of the blood oxygen level-dependent signal time series,the change in the functional connection of the region of interest can be analyzed,so as to distinguish the difference between normal people and patients,and achieve the effect of predicting AD earlier.Studies of work-function connectivity based on resting BOLD signals have been widely used in brain gray matter so far.But in the joint research of white matter,white matter and gray matter,it is not mature yet.Therefore,this article first combines the functional network of the whole brain to partition according to the existing partition template.Then,the mean and variance of the time series of the BOLD signal of the ROI are calculated,and the average time correlation coefficient of the time series is obtained as a feature component reflecting the change of the functional connection.Graph theory and statistical methods were used to analyze the characteristics of normal people,varying degrees of cognitive impairment and functional connectivity in patients with AD.Secondly,the characteristics are constructed and screened by combining the amplitude of low-frequency fluctuation(ALFF)parameter model.Then the similarities and differences in functional connectivity changes are analysed in white matter regions of male and female AD patients by using the two-sample t test.Experiments show that in the resting state,the functional connections in some areas between white and gray matter in AD patients change with the worsening of the disease in AD patients.Male patients with AD have fewer areas of functional connectivity change than women.This further proves that changes in functional connectivity can be used as biomarkers for the diagnosis of AD.Aiming at the change of functional connection that can be used as a biomarker for the diagnosis of AD,this paper proposes to use the combination of convolutional neural network of connection group and dynamic time warping algorithm to extract and classify the functional connection of AD to obtain the best performance feature subset so as to achieve the prediction and diagnosis of AD.First,the classification performance of the classifier with simple neural network and complex neural network is compared.Experiments show that the performance of the connected convolutional neural network classifier and the complex neural network classifier is better,and the performance of the connected convolutional neural network is better.Secondly,the dynamic time warping algorithm is used to describe the functional connection changes of AD,and two feature subsets of warping distance and warping road length are calculated.Correlation coefficient,warped path distance,warped path length,warped distance and warped path combined feature sets are respectively combined with simple neural network classifiers,complex neural network classifiers,and convolutional neural network classifiers to do the feature training and classification effects comparison.The experiments show that the classification accuracy of the combined feature set is high,and the combination of the combined parameters with the convolutional neural network of the connected group achieves the best performance.