Characteristic Analysis Of Human Brain And Assisted Diagnosis Of Alzheimer’s Disease Based On MRI

Medical Imaging Science is playing a more and more important role in medical science research and clinical medicine nowadays. As a relatively advanced nuclear medicine imaging technology for clinical diagnosis, Diffusion Tensor Imaging (DTI) is attracting more attention from domestic and foreign experts. Diagnosis of Alzheimer’s Disease (AD) using DTI has high specificity and safety, and other prominent advantages. For this reason, the characterizing hemispheric asymmetry of T1 and DTI Images in Alzheimer’s Disease is studied in this thesis.A method for characterizing hemispheric asymmetry of T1 and DTI Images in Alzheimer’s Disease is proposed in this thesis, which is designed to help AD diagnosis. All data used in this thesis are selected from the ICBM and ADNI database, including 100 subjects aging from 55 to 90, with both their T1 and DTI images.Since the performance of registration is crucial to the integrity and accuracy of feature extraction, choosing a proper registration algorithm is extremely important for image registration. In this thesis, an affine registration is first applied as a linear model to remove global transformations between images. Once the linear is done, the size, position and direction of aligned images become consistent. Image registration lays the foundation of the following segmentation and feature extraction steps.The Surface Area, Thickness, Mean Curvature and White Matter Volume of T1 images are extracted as features in the thesis. Apparent Diffusion Coefficient, Fractional Anisotropy, Volume Ratio and Relative Anisotropy of DTI are also extracted as features in the thesis. Since T1 image presents more structural information than DTI does, the functional region is segmented from the T1 image, generating a mask. The functional region of each DTI image is obtained using these masks.For characteristic analysis, 2-sample T test, multivariate analysis of covariance (MANCOVA) and analysis of variance (ANOVA) are introduced in this thesis. Then the results are compared and analyzed.With the above methods, theoretical analysis and experiment demonstrate that the proposed method of characterizing hemispheric asymmetry of T1 and DTI Images can assist the diagnosis for Alzheimer’s Disease.