Research On Early Recognition Of Alzheimer’s Disease Based On Radiomics

As an irreversible neurodegenerative disease,Alzheimer’s Disease(AD)is mainly manifested as memory decline and cognitive decline.The early stage of AD is similar to that of the normal aging of the brain.With the development of the disease,the brain structure shrinks mainly in the medial temporal lobe and hippocampus,thus leading to the decline and gradual loss of memory,language,calculation,emotion,and other functions,and seriously affecting the normal life of the patient.So far,effective cure for the disease has not been found.Mild cognitive impairment(MCI),especially amnestic mild cognitive impairment,is a high-risk group of AD.The life and behavioral abilities of MCI patients are not significantly affected,but the probability of conversion to AD is significantly higher than patients with normal control.If the high-risk groups can be accurately identified early,the progress of the disease can be greatly delayed.However,AD is still a degenerative disease with an unclear development process and potential etiology,and there is a lack of biomarkers that can stably characterize AD brain structure and function abnormalities.Therefore,the exploration on robust biomarkers is of great clinical value for the early prevention and treatment of AD.Researchers at home and abroad have carried out a lot of research work on AD.Studies have shown that Amyloid β-protein(Aβ)amyloid deposition and neurofibrillary tangles are the two most important neurobiological markers of AD.The most obvious biological features include atrophy of brain regions such as the medial temporal lobe and hippocampus.Although large-scale brain tissue atrophy can be detected by structural magnetic resonance imaging,there will be some microscopic changes before obvious deformation of brain tissue,which cannot be detected by conventional imaging techniques.Radiomics is proposed as an image analysis method that can obtain more invisible information in images.The radiomics feature of the hippocampus has been confirmed as a new biological marker for AD.However,the brain lesions of AD patients are not limited to the hippocampus.Whether the radiomics features of the whole brain can make a more effective early diagnosis of AD remains to be further explored.In addition,the whole brain constitutes a complex brain network,and the related information needs to be further explored.Based on the above research background,this paper aims to explore whether the brain network topology based on resting-state functional magnetic resonance imaging and the radionics features of the whole brain region based on structural magnetic resonance imaging can be used as potential biomarkers of AD.This study mainly includes two parts: the first work is related to the study of brain network topology based on resting-state functional magnetic resonance imaging,so as to explore the high-risk brain areas of AD.In this paper,a brain network is constructed based on the functional connection matrix of the brain and proposes a brain network topology feature extraction method based on a Deep Walk algorithm.This method characterizes the topological structure of functional connections between brain regions in a more intuitive way.Through the analysis on the differences between AD and normal control(NC)groups,it is found that topological structure features of some brain regions have undergone significant changes,such as Middle Frontal Gyrus,Inferior Frontal Gyrus,Precentral Gyrus,Superior Temporal Gyrus,Middle Temporal Gyrus,Inferior Temporal Gyrus,Fusiform Gyrus,Parahippocampal Gyrus,Postcentral Gyrus,Superior Parietal Lobule,Inferior Parietal Lobule,Basal Ganglia,Thalamus,and Basal Ganglia.The Support Vector Machine(SVM)classification model performs classification verification on the brain network topology features of the AD and NC groups,where Accuracy(ACC)=80.36%,Sensitivity(SEN)=73.67%,Specificity(SPE)=87.17%.Studies have shown that the topological structure of the brain network can be used as a potential AD biomarker.The second work intends to explore the related areas in the whole brain,which can be used as new biomarkers of AD based on structural magnetic resonance imaging.It mainly uses the Radiomics method to extract the high-throughput radiomics features in the structural magnetic resonance images of the whole brain,so as to characterize the changes in the tissue structure of AD,MCI,and NC in the whole brain.First of all,the experimental data were preprocessed and the brain regions were segmented.Secondly,the Pyradiomics method was used to extract a total of 1246 features such as intensity features,shape features,texture features,and wavelet features of each brain area.Then,the two-sample T-test method was used to analyze the significance of the radiomics features of the whole brain in AD,NC,and MCI.Finally,the SVM classification method was used to classify and verify the features of radiomics in the whole brain.The study found that the radiomics features of Hippocampus,Amygdala,Middle Frontal Gyrus,Inferior Frontal Gyrus,Precentral Gyrus,Superior Temporal Gyrus,posterior Superior Temporal Sulcus,Inferior Parietal Lobule,Precuneus,Postcentral Gyrus,Insular Lobe,Cingulate Gyrus,Occipital Lobe,Basal Ganglia,Thalamus,and other brain regions can be used as new biomarkers for AD.Combining the two works in this paper,it is found that there are significant differences in brain network topology features and radiomics features in brain regions such as Middle Frontal Gyrus,Inferior Frontal Gyrus,Precentral Gyrus,Superior Temporal Gyrus,Inferior Parietal Lobule,Insular Lobe,Basal Ganglia and Thalamus,and the research conclusions are consistent.