Alzheimer Disease Neural Fingerprint Study Based On Magnetic Resonance Imaging

With the gradual aging of the world,Alzheimer disease(AD)has become one of the most serious diseases that threaten human health,family life quality and social benign development.However,in view of the current Alzheimer's disease can only be delayed and it can not be completely cured,so for the early diagnosis of Alzheimer's disease will become increasingly important.At present,the diagnosis of AD is still dependent on the clinical symptoms and subjective judgment of professional physicians.Magnetic resonance imaging(MRI)technology provides a noninvasive brain disease examination.Although there are some phenomena such as hippocampal volume change in AD patients by MRI,the problem of quantification of AD is not fully solved because of the low precision of segmentation of multi-brain regions.In this study,based on the automatic segmentation of the multi-atlas,the segmentation of the diffusion tensor image is completed and the texture features of the region of interest are studied.The neural fingerprint model of AD is constructed using the selected brain structure and texture features.The future AD research provides new methods and ideas.Because there is no significant lesion area in AD images,it is necessary to study the structure of the brain specifically for quantifying the MR image of AD.Therefore,accurate segmentation of the brain tissue becomes much more important.In order to improve the precision of image segmentation based on multi-map,two pre-selection methods are proposed in the atlas pre-selection of multi-atlas image segmentation in this paper.One is to use the lateral ventricle structural label to map the pre-selection.And the other is to map the structure of four structural tags of the lateral ventricle,white matter,gray matter and cerebrospinal fluid as a new label atlas,and use the Atlas and Image Automatism of Johns Hopkins University Image Center Segmentation method to segment the brain T1 images.Experimental results show that the proposed two pre-selection methods can improve the segmentation accuracy and segmentation time of the image,and laid the foundation for the segmentation of multi-modal parameter images and further construct neural fingerprints.On the basis of the new image pre-selection method,the segmentation result of T1 structure image is mapped to the other multi-modal parameter images of diffusion resonance imaging(DTI),and the segmentation of multi-modal parameter map is realized.In order to quantitatively analyze the multi-modal parameter map of the 28 regions of interest of the patients with AD,we extracted 13 texture features and used the filtering method to analyze the characteristics of AD.Finally,10 features and 23 regions of interest were identified according to the three types of multi-modal parameter images,and the AD neural fingerprint model with significant significance was constructed.The results show that the proposed method can improve the segmentation accuracy of multi-atlas MR images effectively,and verify the feasibility of constructing neural fingerprints by extracting texture features with multimodal parameter maps.Laid the scientific basis for more comprehensive research in AD areas.