Classification of sMRI data in Alzheimer's disease based on UMPCA and Laplacian Score

Classification of structural magnetic resonance imaging (sMRI) brain scans is helpful to detect Alzheimer's disease (AD) at its early stage. In this paper we present a classification scheme that combines the uncorrelated multilinear principal component analysis (UMPCA) and Laplacian Score (LS) methods, which are known to be effective to the structural correlation preserving and redundancy reduction of the AD-related features hidden in the sMRI. In this scheme, UMPCA is first employed to extract features directly from the tensorial sMRI data of AD subjects and healthy control (HC) subjects. Then, Laplacian Score is used to select the more discriminative features by evaluating their power of locality preserving. Finally, an SVM classifier is built to distinguish AD patients from HC subjects. Experimental results demonstrate that the UMPCA-LS-based method achieves higher recognition accuracy than existing methods.