CLASSIFICATION OF AMNESTIC MILD COGNITIVE IMPAIRMENT USING FMRI

In this work, the feasibility of classifying amnestic mild cognitive impairment (aMCI), a prodromal stage of Alzheimer's disease, was investigated using fMRI activation patterns in the medial temporal lobes (MTL). The activation volume or relative activation extent in each of fourteen subregions of the MTL, when subjects were performing memory tasks, served as features for radial basis function networks (RBFN). The prediction performance was assessed among all combinations of subregions in different memory paradigms and contrasts. A high prediction accuracy of 93.75% (p=2.44x10(-4)) was achieved by as few as two subregions with one contrast. This result demonstrates the possibility of using compact fMRI activation patterns to aid in the diagnosis of aMCI.